Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
A SYSTEM AND METHOD FOR DETECTING PLATELET FUNCTION USING UV LIGHT AND DEEP LEARNING ANALYSES OF MICROSCOPIC IMAGES
Document Type and Number:
WIPO Patent Application WO/2021/105933
Kind Code:
A1
Abstract:
The present invention provides an agonist free system and method for detecting platelet function using UV light, using UV light as stimulant for platelet aggregation, and image analyses of microscopic images of UV treated platelets (sample) using Multi-scale Fully Convolutional Network (MFCN), and training of a deep neural network.

Inventors:
DEB SURYYANI (IN)
CHOWDHURY RANJINI (IN)
SADHU ABHISHEK (IN)
CHOWDHURY ANAL ROY (IN)
DASGUPTA ANJAN KUMAR (IN)
DHARA ASHISH KUMAR (IN)
ROY KRISHNENDU (IN)
BHATTACHARYYA MAITREYEE (IN)
CHAKRABARTI AMLAN (IN)
Application Number:
PCT/IB2020/061203
Publication Date:
June 03, 2021
Filing Date:
November 27, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
SANGUTECH ENTERPRISES PRIVATE LTD (IN)
International Classes:
A61N5/06; G01N15/05; G01N33/49; G06N3/02
Foreign References:
US6043871A2000-03-28
US4066360A1978-01-03
Attorney, Agent or Firm:
MADHAVI, M. (IN)
Download PDF:
Claims:
CLAIMS

We claim:

1. An agonist-free method for detection of platelet function by stimulation of platelet aggregation comprising the steps of: a) collecting whole blood sample and preparing platelet-rich-plasma (PRP) sample; b) stimulating the PRP sample with UV irradiation with wavelength under constant stirring conditions of around 900 to 1200 rpm, more specifically 1000 rpm, at temperature around 30-40 °C, more specifically around 37 °C for 0-10 minutes to obtain UV stimulated PRP; c) smearing UV stimulated PRP on a microscopic slide for microscopic observation ofUV stimulated platelets; d) image acquisition of UV stimulated platelets using an optical microscopic system integrated with a means to capture images; and e) analyzing the images of UV stimulated platelet by implementing multi-scale fully convolutional neural network (MFCN) to distinguish aggregation and non-aggregation of platelets in the image to determine the functionality of the platelets; characterized in that, the UV light is used for stimulating platelets in PRP for platelet aggregation to detect functionality of platelets; and the wavelength of UV used for stimulating PRP is in the range of 240- 310 nm, more specifically 250-260 nm.

2. The agonist-free method for detection of platelet function by stimulation of platelet aggregation as claimed in claim 1, wherein, UV light is used for stimulating platelets in PRP for platelet aggregation to detect functionality of platelets.

3. The agonist-free method for detection of platelet function by stimulation of platelet aggregation as claimed in claim 1, wherein, the wavelength of UV used for stimulating PRP is in the range of 240-310 nm, more specifically 250-260 nm, with maximum UV irradiance optimized at 0.088mW/cm2.

4. The agonist-free method for detection of platelet function by stimulation of platelet aggregation as claimed in claim 1, wherein, analyses of images of UV stimulated platelets in PRP samples by implementing multi-scale fully convolutional neural network (MFCN) comprising steps of: segmenting platelet images using MFCN, wherein, plurality of training images or reference images are created by initial segmentation of platelet images using MFCN by taking an entire image as input and giving dense segmentation and the proposed MFCN consists of six blocks for extraction of multi-scale features, each of the blocks has three convolution layers, the size of the convolution kernel is fixed as 3x3 in all these convolution layers, and dilated convolution is used to preserve the resolution of feature maps and enlarge the receptive field to incorporate larger contextual information, and the images resized to 512x512 pixels to reduce the time of segmentation; determining/evaluating platelet image by features including but not limited to platelet count, free space between the platelet, area covered by the platelet, and their association with aggregated and non- aggregated platelets;

Overlapping of original images, output segmented images and ground truth images for checking the accuracy for detecting platelets, overlap measure is used for checking the accuracy of segmentation, and morphological opening with lxl structural element is used to remove noise from the output segmented images; and training the convolutional neural network for analyze platelet functionality based on platelet aggregation.

5. An agonist-free detection system of platelet function by stimulation of platelet aggregation comprising of: a UV stimulation unit (102) for irradiating PRP with UV light for stimulating platelet aggregation comprising a cabinet (208) with a sliding door (210) with UV protection shield, a sample holding unit (204) having means for temperature variation, a stirring unit (202) beneath the sample holding unit (204), means for irradiating sample with UV light (206), and a controlling unit (212) for maintaining temperature, stirring speed, controlling UV light for irradiation, and time set up; a microscopic system (110) comprising of at least one means of light source, one means to stage a sample, more specifically the UV stimulated PRP, under at least one objective lens, at least one objective lens to focus light from light source onto the sample, series of beam splitters, polarisers, analysers, and mirrors to focus light transmitted/reflected from the sample to an ocular lens; a means for transferring UV stimulated platelets in PRP sample either manually, mechanically, robotically, or automatically from the UV stimulation unit to the microscopic system; a means for image acquisition (112), to capture images with at least 512 x 512 pixels resolution, integrated to the microscopic system; at least one image processing unit for analyzing the images by implementing multi-scale fully convolutional neural network (MFCN) to distinguish aggregation and non-aggregation of platelets in the image to determine the functionality of the platelets; and a means for data storage and transmission to end user via cloud to server; characterized in that, the UV light is used for stimulating platelets in PRP for platelet aggregation to detect functionality of platelets; and the wavelength of UV used for stimulating PRP is in the range of 240-

310 nm, more specifically 250-260 nm.

6. The agonist-free system for detection of platelet function by stimulation of platelet aggregation as claimed in claim 5, wherein, UV light is used for stimulating platelets in PRP for platelet aggregation to detect functionality of platelets.

7. The agonist-free system for detection of platelet function by stimulation of platelet aggregation as claimed in claim 5, wherein, the wavelength of UV used for stimulating PRP is in the range of 240-310 nm, more specifically 250-260 nm, with maximum UV irradiance optimized at 0.088mW/cm2.

8. The agonist-free detection system as claimed in claim 5, wherein, the microscopic system is integrated with the means of image acquisition, preferably a portable magnifying device, micro/mini microscope (110), and means for image acquisition (112) preferably a mobile phone with a camera or a mini camera (112) to capture microscopic images of at least 512x512 pixels resolution.

9. The agonist-free system as claimed in claim 5, wherein, the image processing unit enables analyses of images of UV stimulated platelets in PRP samples by implementing multi-scale fully convolutional neural network (MFCN) comprising steps of: segmenting platelet images using MFCN, wherein, plurality of training images or reference images are created by initial segmentation of platelet images using MFCN by taking an entire image as input and giving dense segmentation and the proposed MFCN consists of six blocks for extraction of multi-scale features, each of the blocks has three convolution layers, the size of the convolution kernel is fixed as

3x3 in all these convolution layers, and dilated convolution is used to preserve the resolution of feature maps and enlarge the receptive field to incorporate larger contextual information, and the images resized to 512x512 pixels to reduce the time of segmentation; determining/evaluating platelet image by features including but not limited to platelet count, free space between the platelet, area covered by the platelet, and their association with aggregated and non- aggregated platelets; overlapping of original images, output segmented images and ground truth images for checking the accuracy for detecting platelets, overlap measure is used for checking the accuracy of segmentation, and morphological opening with lxl structural element is used to remove noise from the output segmented images; and training the convolutional neural network for analyzing platelet functionality based on platelet aggregation.

Description:
A SYSTEM AND METHOD FOR DETECTING PLATELET FUNCTION USING UV LIGHT AND DEEP LEARNING ANALYSES

OF MICROSCOPIC IMAGES

FIELD OF THE INVENTION

This invention relates to a system and method to analyze blood platelet function. More particularly, the present invention relates to an agonist-free system and method to assess platelet function in plasma using ultraviolet (UV) light without any chemical agonists, assisted with deep learning analyses of microscopic images.

BACKGROUND OF THE INVENTION

Platelets are anucleated blood cells of myeloid origin. They are among the smallest corpuscular components of human blood, having a diameter 2-4 pm. The number of platelets in a healthy human typically varies from 150,000/mm 3 to 400,000/mm 3 of blood. The typical shape of resting platelets is discoid, and upon activation they undergo a shape/morphology change to a globular form with pseudopodia (upto 5 pm long) which facilitates the formation of clots.

Circulating platelets are chief effector cells in physiologic hemostasis and pathologic thrombosis, and these are their best-known biologic functions. They circulate in the blood stream as disc-shaped cells that upon activation by blood vessel injury undergo physiological changes that lead to aggregate formation at the site of injury. The aggregate mass evolves from a single platelet to a mass on the order of millimeters in size. The platelet mass additionally recruits and participates with the plasma coagulation proteins leading to the activation of plasma fibrin to form a fibrin clot and stop bleeding.

Altered platelet function is an indicator of several patho-physiological conditions and diseases like, heart attack, cancer, dengue and many more. To elaborate on this, platelet hyper aggregation plays a central role in the pathogenesis of Acute Coronary Syndrome (ACS), commonly known as a heart attack. Anti-platelet drugs are the main of therapy for ACS where the platelet function test is the key diagnosis to test the effectiveness of drugs on platelets. Hemorrhagic dengue, one of the threatening parasitic infection in tropical countries like India, affect platelet number and function, where monitoring of platelets and their function play major role during therapy. Anticancer cytostatic drugs can decrease platelet counts and affect the platelet function which can increase the risk of bleeding and increase the need for platelet transfusion. Reports though showed that severe bleeding episodes occurred (more than 50%) when platelet counts are higher than the recommended platelet counts for transfusion i.e., lOxlOVLacs. This illustrates the urgent need for a method to evaluate instant platelet function and not only the number of circulating platelets.

One of the conventional device/instrument used to quantify blood platelet function are platelet aggrego meters. To determine platelet function, first the blood sample is collected and then the platelet-rich-plasma (PRP) is separated from the other blood cells, i.e. white and red blood cells. This PRP sample is subjected to a specific agonist (activator) for aggregation of platelets and then the sample is subjected to an optical transmittance procedure. The optical transmittance aggregometers are based on the technique of detection of light transmitted through a cuvette containing platelet-rich plasma (PRP). As platelets form aggregates, the light transmission through the blood increases in proportion to the aggregation response. The optical transmittance method attempts to detect the shape change, the rate of aggregation, the size of the aggregates, and the maximum aggregation (represents as % of aggregation) of platelets.

US6043871A discloses an apparatus for assessing the function of platelets in a sample of whole blood measuring the platelet aggregation in whole blood in response to standard aggregating agents, i.e. agonist. The measurement is based on impedance method and does not require separation of erythrocytes from blood but need expensive electrodes. However, the method uses expensive aggregating agent selected from the group consisting of: ADP, collagen, ristocetin, and epinephrine, and requires expensive electrodes.

Till date all the established methods for the platelet function study (for example, optical and whole blood Platelet aggregometry, VerifyNow, Plateletworks, Thrombelastograph, Platelet Mapping System, Impact cone and plate(let) analyzer, PFA-100, VASP phosphorylation state) require recurrent use of agonists (extremely expensive), costly reagents, huge laboratory set up (skilled personnel, air condition machine, computer, 4°C, - 20°C and -80°C refrigerators etc.), big space etc. which are not compatible and achievable in remote health centers and diagnostic centers. The major issue is the use of expensive agonists such as adenosine diphosphate (ADP), collagen, thrombin, arachidonic acid, and thromboxane A2, and others. Another problem is relying of spectrophotometric method and/or impedance for determining aggregation which makes the device less compatible in size and cost for use as a point-of-care device.

Therefore, it is the need of the hour for a cost-effective, easy accessible and telecommunication enabled technique to measure instant platelet function especially in a country like India where population, communication and monetary constraint are the major issues.

Hence it is not only important to create a system for an expeditious analysis but it is also important to make the system smart which increases its efficiency with each exposure and becomes self-trained. Taking into consideration the prior art, the present invention provides a system and method for detecting platelet function using an aggregation stimulant that is inexpensive and reusable, and a system that is fast/speedy /instant, smart, and portable.

OBJECT(S) OF THE INVENTION

The main object of the invention is to provide an agonist free system and method for detecting platelet function using electromagnetic wave as a stimulant for platelet aggregation, more specifically UV light, and analyses of microscopic images of platelet sample implementing Multi-scale Fully Convolutional Neural Network (MFCN) for determining/evaluating diagnostic feature platelet image i.e. platelet function with respect to platelet aggregation in the sample.

Another object of the invention is to provide an agonist free method for detecting platelet function using electromagnetic wave as a stimulant for platelet aggregation, wherein UV light is the stimulant for platelet aggregation.

Yet another object of the invention is to provide a method to analyze aggregation of platelets to determine its functionality microscopically wherein, the method involves analyses of microscopic images of platelet sample implementing MFCN for determining/evaluating diagnostic feature platelet image i.e. platelet function with respect to platelet aggregation in the sample by platelet count, free space between the platelet, and area covered by the platelet and their association with aggregated and non-aggregated platelets.

Yet another object of the invention is to provide a system for detecting platelet function using electromagnetic wave, more specifically UV light, wherein, the system is easily portable and is a point-of-care solution which is enabled by using simple portable micro/mini microscope and a mini camera or mobile phone enabled camera for image capturing.

SUMMARY OF THE INVENTION

In carrying out the above objects and other objects of the present invention, the present invention provides an agonist free system and method for detecting platelet function using UV light as stimulant of platelet aggregation, and analyses of microscopic images of platelet sample implementing MFCN for determining/ evaluating diagnostic feature platelet image i.e. platelet function with respect to platelet aggregation in the sample by platelet count, free space between the platelet, and area covered by the platelet and their association with aggregated and non-aggregated platelets, and training of a deep neural network. This system provides a method to detect platelet function without any use of expensive chemicals or biological agonists which provides a point-of-care solution enabling operations in remote locations in a simple and inexpensive set-up. Most of the established methods for platelet function assessment rely on impedance and optical methods to determine platelet aggregation, whereas, the present invention provides a method to determine platelet aggregation microscopically without use of any chemical agonist.

In one embodiment the invention provides a method for detecting platelet function comprising of: collecting whole blood sample and preparing platelet-rich-plasma (PRP) sample from the whole blood sample, at temperature around 20-40 °C; stimulating the PRP sample with UV irradiation to obtain UV stimulated PRP, wherein, irradiation of the PRP sample with UV acts as a stimulator for platelet aggregation, the PRP sample is irradiated with at least one UV wavelength of 240-310 nm under constant stirring conditions of around 900 to 1200 rpm more specifically 1000 rpm and at temperature around 30-40 °C, more specifically around 37 °C to maintain human body ambient temperature condition for 0-10 minutes; smearing UV stimulated PRP on a microscopic slide for microscopic observation ofUV stimulated platelets image acquisition of UV stimulated platelets using an optical microscopic system integrated with a means to capture images; and analysing the images for platelet aggregation, wherein, the captured images are segmented using MFCN, which takes entire image as input and give a dense segmentation, the parameters such as count of platelets, free space and area/size of platelets are used to distinguish aggregation and non-aggregation of platelets in the image to determine the functionality of the platelets.

The method further comprises the step of training the network in the cloud. The term cloud also referred to as cloud computing is intended to mean one or more computer networks, server(s), storage(s), application(s) and/or service(s) that is/are remotely accessible by users. For training the network in the cloud, training data comprising microscopic images are uploaded into the server through cloud and a network is trained by the microscopic images. The uploading can take place by any means of data transfer such as by cables, wireless and/or both and can be done sequentially, in packages and/or in parallel.

In another embodiment, the invention provides a system for agonist free detection of platelet function of a sample, wherein, the system comprises of: a UV stimulation unit (102) for irradiating PRP with UV light for stimulating platelet aggregation comprising a cabinet (208) with a sliding door (210) with UV protection shield, a sample holding unit (204) having means for temperature variation, a stirring unit (202) beneath the sample holding unit (204), means for irradiating sample with UV light (206), and a controlling unit (212) for maintaining temperature, stirring speed, controlling UV light for irradiation, and time set up; a microscopic system (110) comprising of at least one means of light source, one means to stage a sample, more specifically the UV stimulated PRP, under at least one objective lens, at least one objective lens to focus light from light source onto the sample, series of beam splitters, polarisers, analysers, and mirrors to focus light transmitted/ reflected from the sample to an ocular lens; a means for transferring UV stimulated platelets in PRP sample either manually, mechanically, robotically, or automatically from the UV stimulation unit to the microscopic system; a means for image acquisition (112), to capture images with at least 512 x 512 pixels resolution, integrated to the microscopic system; at least one image processing unit for analyzing the images by implementing multi-scale fully convolutional neural network (MFCN) to distinguish aggregation and non-aggregation of platelets in the image to determine the functionality of the platelets; and a means for data storage and transmission to end user via cloud to server.

The invention provides a system which utilizes UV light for stimulating platelets in PRP for platelet aggregation to detect functionality of platelets, and the wavelength of UV used for stimulating PRP is in the range of 240-310 nm, more specifically 250-260 nm.

The invention provides a microscopic system and means for image acquisition (106) the microscopic system (110) integrated to a means for capturing images (112) for capturing microscopic images of UV stimulated platelets in PRP. The microscopic system (110) comprises of at least one means of light source, one means to stage a sample, more specifically the UV stimulated PRP, under at least one objective lens, at least one objective lens to focus light from light source onto the sample, series of beam splitters, polarisers, analysers, and mirrors to focus light transmitted/reflected from the sample to an ocular lens, and the ocular lens integrated to the means to capture images (112) as observed from the ocular lens, and the images of at least 512 x 512 pixels are acquired.

The invention further provides at least one image processing unit (108) which enables analyses of images of UV stimulated platelets in PRP samples by implementing multi-scale fully convolutional neural network (MFCN) comprising steps of: segmenting platelet PRP sample images using MFCN, wherein, plurality of training images or reference images are created by initial segmentation of platelet PRP images using MFCN by taking an entire image as input and giving dense segmentation and the proposed MFCN consists of six blocks for extraction of multi-scale features, each of the blocks has three convolution layers, the size of the convolution kernel is fixed as 3x3 in all these convolution layers, and dilated convolution is used to preserve the resolution of feature maps and enlarge the receptive field to incorporate larger contextual information, and the images resized to 512x512 pixels to reduce the time of segmentation; determining/evaluating platelet image by features including but not limited to platelet count, free space between the platelet, area covered by the platelet, and their association with aggregated and non-aggregated platelets; overlapping of original images, output segmented images and ground truth images for checking the accuracy for detecting platelets, overlap measure is used for checking the accuracy of segmentation, and morphological opening with lxl structural element is used to remove noise from the output segmented images; and training the convolutional neural network for analyzing platelet functionality based on platelet aggregation.

The invention also provides a means for data storage and transmission to end user via cloud to server.

In yet another embodiment, the invention provides a mobile application based platform for registering for the platelet function test, and uploading/ downloading and storage of images and platelet function test results accessible to parties requesting for platelet function test and parties conducting platelet function test. This enables remote access of platelet function test images and results.

BRIEF DESCRIPTION OF THE DRAWING

The object of the invention may be understood in more details and more particularly description of the invention briefly summarized above by reference to certain embodiments thereof which are illustrated in the appended drawings, which drawings form a part of this specification. It is to be noted, however, that the appended drawings illustrate preferred embodiments of the invention and are therefore not to be considered limiting of its scope, for the invention may admit to other equally effective equivalent embodiments.

Fig. 1 depicts a schematic outlining the system (100) for detecting platelet function based on its aggregation properties upon stimuli with UV irradiation; Fig. 2 depicts a UV stimulation unit (102);

Fig. 3a depicts a microscopic system and means for image acquisition (106) comprising a portable micro/mini microscope (110), a mobile phone with a camera (112), means to hold the UV stimulated PRP sample over the microscope (302), and means to hold the mobile phone with a camera (304) over the portable micro/mini microscope (110);

Fig. 3b depicts a microscopic system and means for image acquisition (106) comprising a portable micro/mini microscope (110), a mini camera (112), means to hold the UV stimulated PRP sample under the microscope (302), means to hold the mini camera (304) over the portable micro/mini microscope (110), and a light source for the mini microscope (306) to irradiate the UV stimulated PRP sample for capturing images;

Fig. 4 depicts a workflow of image analysis;

Fig. 5 depicts architecture of Multi-scale Fully Convolutional Neural Network (MFCN);

Fig. 6a depicts the important parameters to distinguish aggregated platelets through MFCN based image analysis;

Fig. 6b depicts the important parameters to distinguish non-aggregated platelets through MFCN based image analysis;

Fig. 7a is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins w.r.t. number of counts/objects in the image, where N=5 i.e number of experiments is 5;

Fig. 7b is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins w.r.t. free space in the image where N=5 i.e number of experiments is 5;

Fig. 7c is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins w.r.t area vs counts in the image where N=5 i.e number of experiments is 5;

Fig. 8a is a graphical representation of effect of UV irradiation of platelets at wavelengths ranging from 240 nm to 300 nm for 10 minutes w.r.t. number of counts/objects in the image where N=5 i.e number of experiments is 5;

Fig. 8b is a graphical representation of effect of UV irradiation of platelets at wavelengths ranging from 240 nm to 300 nm for 10 minutes w.r.t. free space in the image where N=5 i.e number of experiments is 5;

Fig. 9a is a graphical representation of effect of UV irradiation of platelets at 260 nm for varying time periods to a maximum of 10 minutes w.r.t. number of counts/objects in the image;

Fig. 9b is a graphical representation of effect of UV irradiation of platelets at 260 nm for varying time periods to a maximum of 10 minutes w.r.t. free space in the image;

Fig. 10a is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins with or without treatment with Asprin w.r.t. number of counts/objects in the image, where N=5 i.e number of experiments is 5;

Fig. 10b is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins with or without treatment with Asprin w.r.t. free space in the image, where N=5 i.e number of experiments is 5;

Fig. 10c is a graphical representation of effect of UV irradiation of platelets at 260 nm for 10 mins with or without treatment with Asprin w.r.t. area vs count in the image; and

Fig. 11 depicts a) and (d) are input raw images; (b) and (e) are initial segmentation results of MFCN; (c) and (f) are final segmentation obtained after morphological operations. DESCRIPTION OF THE INVENTION

The present invention will now be described more fully hereinafter with reference to the accompanying drawings in which a preferred embodiment of the invention is shown. This invention may, however, be embodied in many different forms and should not be construed as being limited to the embodiment set forth herein. Rather, the embodiment is provided so that this disclosure will be thorough, and will fully convey the scope of the invention to those skilled in the art.

The present invention relates to an agonist free system and method for detecting platelet function using UV light as stimulant of platelet aggregation, and analyses of microscopic images of platelet sample implementing MFCN for determining/ evaluating diagnostic feature platelet image i.e. platelet function with respect to platelet aggregation in the sample by platelet count, free space between the platelet, and area covered by the platelet and their association with aggregated and non-aggregated platelets, and training of a deep neural network. This system provides a method to detect platelet function without any use of expensive chemicals or biological agonists which provides a point-of-care solution enabling operations in remote locations in a simple and inexpensive set-up.

EXAMPLE 1

SYSTEM FOR DETECTING PLATELET FUNCTION Fig. 1 provides a schematic outlining the system (100) for detecting platelet function based on its aggregation properties, wherein the system comprises of a UV stimulation unit (102) for irradiating PRP with UV light for stimulating platelet aggregation, means for microscopic system and image acquisition (106), means for transferring UV stimulated platelet sample (104) either manually, mechanically, robotically, or automatically from the UV stimulation unit (102) to microscopic system and means for image acquisition (106), and image processing unit (108). As depicted in Fig. 2 the UV stimulation unit (102) further comprises of cabinet (208) with a sliding door (210) with UV protection shield, a sample holding unit (204) having means for temperature variation, a stirring unit (202) beneath the sample holding unit (204), means for irradiating sample with UV light (206), and a controlling unit (212) for maintaining temperature, stirring speed, controlling UV light for irradiation, and time set up. The means for microscopic system and image acquisition (106) further comprises of a microscopic system (110) comprising of at least one means of light source, one means to stage a sample, more specifically the UV stimulated PRP, under at least one objective lens, at least one objective lens to focus light from light source onto the sample, series of beam splitters, polarisers, analysers, and mirrors to focus light transmitted/reflected from the sample to an ocular lens, and the ocular lens integrated to the means to capture images (112) as observed from the ocular lens. The image processing unit (108), wherein the analyses of platelet images is done by implementing MFCN, and parameters such as count, free space and area/size of platelets are used to distinguish aggregation and non aggregation of platelets in the image to determine the functionality of the platelets. In an alternate embodiment, as depicted in Fig. 3a the invention provides a system for detecting platelet function based on its aggregation properties, wherein the microscopic system and means for image acquisition (106) is more portable comprising of a portable micro/mini microscope (110), a mobile phone with a camera (112) to capture microscopic images, means to hold the UV stimulated PRP sample onto the microscope (302) for observation and capturing images of UV stimulated platelets in the sample, and means to hold the mobile phone with a camera (304) over the portable micro/mini microscope (110). This makes the invention simple, point-to-care facility which can be easily set-up at remote locations. In yet another alternate embodiment, as depicted in Fig. 3b the invention provides a system for detecting platelet function based on its aggregation properties, wherein the microscopic system and means for image acquisition (106) is portable comprising of a portable micro/mini microscope (110), a mini camera (112) to capture microscopic images, means to hold the UV stimulated PRP sample under the microscope (302) for observation and capturing images, means to hold the mini camera (304) over the portable micro/mini microscope (110), and a light source for the mini microscope (306) to irradiate the UV stimulated PRP sample for capturing images. This makes the invention simple, point-to-care facility which can be easily set-up at remote locations.

One of the optimized systems has the components as listed in Table 1.

Table 1: Components of an optimized system

EXAMPLE 2

METHOD FOR PRP SAMPLE PREPARATION AND IMAGE

ACQUISITION

I. Isolation of Platelet Rich Plasma(PRP)

9ml of blood was drawn using 10 ml syringe (22G X 1.25 inch) from voluntary subjects with their due permission and consent (ethical clearance number MC/KOL/IEC/NON-SPON/136/09-2018). Blood was then mixed with 1ml of 3.2% tri-sodium citrate (an anticoagulant). The citrated blood was allowed to stand for 10 minutes before centrifugation at 200g (1260-1300 rpm) for 10 minutes. The whole process was done at room temperature. Upper two third portion of the supernatant obtained from the centrifuged citrated blood was carefully collected into another tube as the PRP.

II. UV stimulation of PRP lml of PRP taken in a quartz/siliconized glass cuvette (Hellma Fluorescence quartz cuvette) was exposed under UV radiation (from a wavelength range of 240 nm to 310 nm of with each interval of lOnm). The temperature is firmly maintained at 37°C, to maintain the normal physiologic body temperature even in external environment condition. The sample is kept under a constant stirring condition of around 1000 rpm, this is because inside the body the blood is always in a flowing state and never static (unless it clots). The highest irradiance of the UV exposure was 0.088mW/cm 2 .

III. Image acquisition procedure:

UV light stimulated PRP samples (20m1) were taken at regular intervals of time (0 minutes, 5 minutes & lOminutes) and smeared on a clean microscopic glass slide (Microscope Slides, Ground Edges, RIVIERA Size75 x 25 mm) to observe UV light stimulated platelets in the sample under a simple light microscope. The maximum time for which the platelets were UV stimulated was of 10 minutes. Images at 10X and 40X magnification were observed under microscope (Nikon Eclipse TiU, Japan) and captured through microscope linked camera (Nikon digital sight DS-Ril). The resolution of the microscopic images was 1280 x 1024 pixels.

EXAMPLE 3

METHOD FOR IMAGE ANALYSES

One of the methods for analyzing an image is use of multi-scale fully convolutional neural networks (MFCN) which is a powerful tool for segmentation of different morphological regions. MFCNs owe their name to their architecture, which is built only from locally connected layers, such as convolution, pooling and up-sampling etc. There is no dense layer used in this kind of architecture. This reduces the number of parameters and computation time. Also, the network can work regardless of the original image size, without requiring any fixed number of units at any stage, given that all connections are local. Compared with classification and detection tasks, segmentation is a much more difficult task. Important tasks in are:

Image Classification: Classify the object (Recognize the object class) within an image;

Object Detection: Classify and detect the object(s) within an image with bounding box(es) bounded the object(s), which means one also needs to know the class, position and size of each object; and

Semantic Segmentation: Classify the object class for each pixel within an image, which means there is a label for each pixel.

To obtain a segmentation map (output), segmentation networks usually have 2 parts:

Down sampling path: capture semantic/contextual information; and Up sampling path: recover spatial information.

The downsampling path is used to extract and interpret the context (what), while the upsampling path is used to enable precise localization (where). Furthermore, to fully recover the fine-grained spatial information lost in the pooling or downsampling layers, one often uses skip connections. A skip connection is a connection that bypasses at least one layer. Here, it is often used to transfer local information by concatenating or summing feature maps from the downsampling path with feature maps from the upsampling path. Merging features from various resolution levels helps to combine context information with spatial information.

The images for platelet aggregation was analysed, wherein, the captured images are segmented using MFCN as shown in Fig. 4.

Accurate segmentation of platelet image is important for quantitative analysis of platelets aggregation and platelet count. Poor contrast between the platelets and its background, ill-defined boundary of platelets, and similar appearance of platelets clusters and aggregate are the major challenges for acceptable segmentation. Traditional image segmentation techniques such as thresholding, region growing, edge detection and level set method are unable to segment the platelet image with sufficient accuracy. Therefore, MFCN has been used for image segmentation, which takes an entire image as input and give a dense segmentation as shown in Fig. 5. MFCN is a powerful technique for segmentation of different morphological regions. MFCN is built from locally connected layers using only convolution operation. Each blocks of MFCN has three convolution layers; the size of the convolution kernel is fixed as 3x3 in all these convolution layers. The number of filter in each layer is 64. Dilated convolution is used to preserve the resolution of feature maps and enlarge the receptive field to incorporate larger contextual information. The number of trainable parameters is sufficiently low in MFCN. Therefore, time of segmentation is very small. The network can work regardless of the original image size.

The ground truth of a few images was created by experienced pathologist for training of a deep neural network. The images were resized to 512x512 pixels to reduce the time of segmentation. Several augmentation techniques such as flip and rotation were performed to increase number of training images. A stochastic gradient-based optimization ADAM was applied to minimize the cross-entropy based cost function. The learning rate for the ADAM optimizer was set to 0.0001 and over-fitting was reduced by using dropout. The weights of background and foreground were maintained as 1:10 and training were performed up to 20 epochs. The hyper-parameters were determined based on the validation dataset. Morphological opening (with lxl structural elements) was performed to remove noise in the initial segmented image. The count of platelets, free space and area of platelets are considered as diagnostic feature to distinguish aggregation and non-aggregation of platelets in the image as well as to determine the functionality of the platelets.

EXAMPLE 4 RESULTS

As shown in Fig. 6, the parameters used for image analyses are such as count, free space and area/size of platelets are used to distinguish aggregation and non-aggregation of platelets in the image to determine the functionality of the platelets. As depicted in Fig. 6a, the normal platelets get aggregated upon UV irradiation and as depicted in Fig. 6b, the anti-platelet drug treated platelets do not aggregate.

As depicted in the graph of Fig. 7a, the UV irradiation of normal platelets decreases the count/number of objects in the image with increase in time from 0 minutes to 10 minutes because of clustering of platelets into aggregates. Similarly, as depicted in the graph of Fig. 7b, the free space in the image increases with increase in time from 0 minutes to 10 minutes because of platelet aggregation upon UV irradiation. As depicted in the graph Fig. 7c, there are a greater number of counts in dispersed area at 0 minute UV irradiation whereas there are less number of counts and restricted to less area upon 10 minute irradiation suggesting aggregation.

I. Optimized UV wavelength for platelet aggregation:

Experiments were conducted to determine the wavelength frequency in the UV range of 240nm to 310nm which caused maximum platelet aggregation within 10 minutes. As shown in Fig. 8, the dominant wavelength in which the maximum aggregation, leading decrease in count and increase in free space, was observed was in the range of 250nm to 260 nm.

II. Optimised time of UV exposure and irradiance:

Experiments were conducted to determine the optimum time for maximum platelet aggregation at the optimized wavelength 260 nm. As per the graphical representations in Fig. 9, the time kinetics showed the maximum aggregation was attended after 10 minutes. The maximum UV irradiance was also optimized at 0.088 mW/cm2.

III. Effect of anti-platelet drug Asprin on UV stimulation of platelets:

To understand the difference between aggregation of platelets and non aggregation of platelets using the present system and method, the effect of the antiplatelet drugs on the platelet aggregation using UV as stimulant was assessed. Most widely used antiplatelet drug aspirin (Acetyl Salicylic Acid SIGMA: A2093-100G) were incubated with the PRP sample to its final concentration of 3mM for 30 minutes at 37°C. After incubation the aspirin incubated sample were exposed to UV radiation upto 10-15 minutes. Hence, normal PRP when stimulated should show aggregation properties upon stimulus such as UV and others, and PRP treated with anti-platelet drug should inhibit platelet aggregation upon stimulus such as UV and others as the anti-platelet drug inhibit platelets function.

From the graphical representations depicted in Fig. 10, it is clear that in presence of aspirin (3mM) platelets’ aggregation decreased when exposed to UV (260nm) compared to the control set (where 260nm UV exposure carried out without aspirin). Hence, it can be said that the UV is acting as a natural agonist as clinically used drug aspirin was able to reduce the UV induced platelet aggregation.

IV. Detection of platelet aggregation using deep learning or neural network:

The overlap (original & segmented) images obtained via deep learning (as explained in Example 3 are shown in Fig 11. Morphological opening (with lxl structural element) is performed to remove noise in the initial segmented image.

Fig. ll(a)and(d) are input raw images of aggregated and non-aggregated platelet images respectively; Fig. 11 (b)and(e) are initial segmentation results of MFCN; Fig. 11 (c)and (f) are final segmentation obtain after morphological opening of respective images.

While certain exemplary embodiments have been described and shown in the accompanying drawings, it is to be understood that such embodiments are merely illustrative of, and not restrictive on, the broad invention, and that this invention not be limited to the specific constructions and arrangements shown and described, since various other changes, combinations, omissions, modifications and substitutions, in addition to those set forth in the above paragraphs, are possible. Those skilled in the art will appreciate that various adaptations and modifications of the just described embodiments can be configured without departing from the scope and spirit of the invention. Therefore, it is to be understood that, within the scope of the appended claims, the invention may be practiced other than as specifically described herein.