DESCRIPTION

Field of the invention

In the most general aspect thereof, the present invention relates to the field of intraoperative radiotherapy and the processing of personalized treatment plans consistent with relevant criticalities and provides a computed method which allows obtaining, without resorting to traditional imaging techniques and therefore saving time, resources, and patient radiation dose, an intraoperative computed tomography or CT image which is representative of the intraoperative condition of the operated subject with sufficient precision for the purposes of processing a treatment plan or TPS (Treatment Planning System) in IORT.

Prior art

IORT (Intra Operative Radiation Therapy) is a form of radiotherapy carried out directly in the operating room, at the end of the tumor removal operation itself. The main advantage with respect to “conventional radiotherapy” (i.e., “with external beams”) is the possibility of delivering the radiation dose (electrons or X-rays) selectively to the target of interest. For example, in the case of breast cancer, it is possible to treat the tumor bed at the end of the removal, with the aim of killing any micro-metastases which are not visible/surgically removable. Furthermore, if required, given the surgical context it is also possible to physically move/shield healthy organs at risk which must be protected from the radiation.

The first uses of the technique date back to the 1960s when, however, the treatment required moving patients from the operating room to the accelerator complex, with obvious logistical difficulties. Therefore, the true diffusion of IORT only occurred in the 1990s, by virtue of the development of “mobile” accelerators which could be brought to the operating room at the time of treatment. This considerable logistical simplification made the advantages in terms of dose conformation offered by IORT even more attractive.

The choice of the radiotherapy treatment parameters such as dose and energy of the particles to be used (i.e., the depth to be treated) is still carried out based on general tables, in stark contrast to the principle of personalized medicine, the true paradigm of modern medicine. Furthermore, even once the treatment has been carried out, there is currently no way to provide the patient with the aforesaid “dosimetry report” which takes into account his/her particular treatment case. The combination of all these effects causes IORT to be applied with global uncertainties (e.g., on the deposited dose) of around 10%. The current limitations of IORT stem from the fact that no TPS (Treatment Planning System) has ever really been developed for IORT. The fundamental reason for this shortcoming is the unavailability of intraoperative imaging from which it is possible to obtain the “virtual patient” required for the TPS (Hensley FW. Present state and issues in IORT Physics. Radiation Oncology. 2017).

In fact, while in external beam therapy, a previous CT is considered a good representation of the patient (net of the changes which occurred in ~2 weeks between CT and treatment, alignment problems, breathing problems, etc., which form the systematic uncertainty of the TPS itself), in the case of IORT a preoperative image is of no use for a possible TPS. In fact, at the end of the operation, the patient’s anatomy has undergone very significant changes, especially in the zone where the treatment will be carried out.

There are some attempts in the literature to develop TPS for IORT, which typically follow two approaches. The first approach is to perform a true intraoperative imaging, for example by performing a CT or NMR on the patient at the end of the operation (Garcia-Vazquez V, Marinetto E, Guerra P, Valdivieso-Casique MF, Calvo FA, Alvarado-Vasquez E, et al. Assessment of intraoperative 3D imaging alternatives for IOERT dose estimation. Z Med Phys. 2017). This path, which would also provide an ideal image for TPS, not only encounters logistical difficulties such as having to move the patient “open” from the operating room to the CT and vice versa, but also difficulties of an economic and time nature and regarding the further dosing to the patient, which in fact compromised the true development thereof.

Another path which has been attempted is to modify the patient’s preoperative CT images “ex-post”, for example by “removing” (i.e., “deleting” from the CT) some parts of the body to imitate access to the surgical cavity (Valdivieso-Casique MF, Rodriguez R, Rodriguez- Bescos S, Lardies D, Guerra P, Ledesma MJ, et al. RADIANCE-A planning software for intra-operative radiation therapy. Translational Cancer Research. 2015). The limit of this approach is that the human body is very mobile, and the approximation introduced by a “rigid” removal of the tissues, which does not take into account, for example, the displacements which occur due to the incision of the tissue, is so crude as to introduce unacceptable errors in the simulations.

Therefore, the need is felt for a method for processing treatment plans for IORT which overcomes the drawbacks of the prior art described above and which allows obtaining dosimetry reports with an uncertainty below 10%.

Summary of the Invention

The invention consists of a computed method for obtaining a corrected or intraoperative tomography which corresponds to the state of the subject subjected to surgery. The invention further consists of a method for processing treatment plans in lORT-type operations. The latter method is based on the utilization of the above computed method to obtain a CT image of the patient at the end of the surgery and before the radiotherapy treatment. Such an image is then used as an input for a simulation algorithm of the physical treatment processes, allowing a map of the dose released to each organ to be obtained.

Further features and advantages of the invention will become apparent from the following detailed description with reference to the accompanying drawings.

Brief description of the Figures

Figure 1 is a diagrammatic depiction of a first embodiment of the computed method of the invention.

Figure 2 is a diagrammatic depiction of a second embodiment of the computed method of the invention.

Figure 3 shows a diagrammatic depiction of a use mode of the computed method of the invention for processing treatment plans for IORT operations.

Detailed description of the invention

The present patent application relates to a procedure which allows obtaining a computed tomography or CT image which is representative of the intraoperative condition of the operated subject with sufficient precision for the purposes of processing a treatment plan or TPS (Treatment Planning System) in IORT. The IORT operation generally comprises two main steps which are: the so-called surgery during which the removal of the diseased tissues occurs, and the actual radiotherapy treatment which consists in exposing the surrounding tissues to ionizing radiations: for brevity, these steps will be referred to as “surgery” and “radiotherapy” below, respectively. The hypothesis underlying the present patent application is that during surgery the anatomical zone which undergoes significant changes is substantially limited in extent, while for the remaining part of the patient’s body the preoperative CT continues to remain a good approximation of the actual situation at the end of the surgery. For example, in the case of the removal of breast tumors, only the operated breast undergoes a significant deformation of the anatomy thereof, while the rest of the body remains virtually intact. Preoperative CT or preoperative tomography thus means the computed tomography image of the subject to be operated on generally acquired in the days preceding the surgery. Therefore, based on the hypothesis of a limited variation in the real anatomy of the patient during surgery with respect to the preoperative CT, the invention suggests a computed method for correcting the preoperative CT in which the missing or modified volumetric information is corrected so as to create an “intraoperative CT” which best approximates the subject’s condition at the moment of the radiotherapy treatment or radiotherapy.

According to the invention, the computed method of the present patent application comprises the following main steps: a. Providing a preoperative tomography of the subject and a first three-dimensional scan of the region involved in the operation; b. Obtaining, from said preoperative tomography and first three- dimensional scan, information related to size and spacing of the voxels forming them; c. Calculating the average density of the subject in the region to be subjected to surgery from the preoperative tomography; d. Trying to find the space transformation which maximizes the alignment between said preoperative tomography and first three- dimensional scan; e. Applying said transformation to a second three-dimensional scan of the region involved in the surgery to obtain a three-dimensional scan aligned with the preoperative tomography; f. Correcting the preoperative tomography to obtain an intraoperative computed tomography with the following modes: f1. Assigning to the voxels which are empty in the preoperative tomography and present in the aligned three-dimensional scan a density equal to the average density as per step c; f2. Assigning to the voxels which are present in the preoperative tomography and empty in the aligned three-dimensional scan a density equal to that of air; wherein at least said second three-dimensional scan relates to the region involved in the operation after the surgery.

For the purposes of the present invention, empty voxels means the unit of volume not occupied by tissue of the subject to be subjected to the surgical treatment.

In a preferred embodiment of the invention, said first and second three-dimensional scans of the region involved in the operation coincide and are acquired at the end of the surgery or after the removal of the cancer tissue and in any case before the skin layer is restored. In a variant of the invention, in order to improve the effectiveness of the alignment step of the preoperative tomography and the first three- dimensional scan of the region involved in the operation, said first three- dimensional scan is acquired prior to surgery while the second three- dimensional scan is acquired with reference to the same region at the end of the surgical stage intended as in the previous paragraph. Said first and second three-dimensional scans of the region involved in the operation are preferably acquired by the same acquisition system and with the same position related to the individual to subject/subjected to surgery. The mathematical transformation obtained at the end of step d is always applied to the second three-dimensional scan, whether this coincides with the first scan or not. Step a.

The preoperative CT of the patient (as per step a) can be acquired in the days preceding the treatment, as is normal practice for external beam radiotherapy, but also in the hours immediately preceding it, since the execution times of the correction algorithm of said CT are very fast (~10s). In this latter case, the accuracy of the resulting image with respect to the real morphology of the patient will be even greater.

Step b.

The information related to the size and spacing of the voxels of the two images (preoperative CT and superficial scan) (step b) is necessary to map the two images, coming from very different technologies, on a common space, where the alignment can then be searched for. Such features are easily obtained from both CT images (result of the features of the machinery used and reported in the metadata of the DICOM files) and surface scanning (result of the settings chosen for the scanning device).

Step c.

The evaluation of the average density of the voxels in the zone involved in the surgery or which will be subjected to surgical modification (as per step c) is necessary to minimize the error of the preoperative CT correction procedure. Such an evaluation can be obtained by cropping (both manual and possibly automated, for example by machine learning algorithms) of the zone on the preoperative CT image. Those skilled in the art will be able to determine the extent of said region involved in the surgery in order to obtain a significant average density value for the purposes of the suggested method. Merely by way of example, in the case of an IORT operation on the breast, said region will be limited to the breast itself rather than involving the entire thoracic image of the patient. Step d.

The transformation which maximizes the alignment of the surface scan with the preoperative CT (step d) is identified by means of stochastic algorithms, based for example (but not only) on Gradient Descent. In some cases, in order to maximize the effectiveness of the algorithm for identifying said transformation, it is possible to outline on the preoperative CT image the zone where the surgical modification is expected, in order to exclude it from the area to be searched for the alignment between the 3D scan and said preoperative CT.

Step e.

The application of said transformation to the volumetric scan (step e) leads to obtaining a volumetric scan which is homogeneous in the spatial coordinates and correctly aligned with the preoperative CT.

Step f.

At the end of the correction procedure (step f), a CT image is obtained (therefore in an appropriate format, for example (but not exclusively) DICOM) which is representative of the anatomy modified during surgery (“intraoperative CT”), and which can then be used as an input for simulation algorithms, for example to obtain treatment plans.

Use of the computed method for processing TPS

The computed method suggested is used in a protocol aimed at processing treatment plans for IORT operations. The treatment plan, also known as TPS (Treatment Planning System), consists of a process which involves multiple methods and different professionals, and which allows the accurate planning of an external beam radiotherapy treatment.

The process is based on a morphological image of the patient (typically a CT), with which a “virtual patient” is created, on which the radiotherapist indicates the target of the therapy and the dose to be received by this volume. This virtual patient is then used to perform computer simulations (pencil beam algorithms, Monte Carlo, etc.) which allow comparing various treatment configurations, with the aim of identifying the best compromise in terms of dosing to the tumor/dosing to healthy tissues.

The use of the computed method for processing treatment plans for IORT operations comprises the following main steps:

A. Acquiring a preoperative tomography of the subject;

B. Acquiring a three-dimensional scan of the region involved in the operation under the conditions recorded at the end of the surgery;

C. Performing the computed method in the steps a-f thereof to obtain an intraoperative computed tomography;

D. Simulating the radiation therapy by simulation algorithms based on the intraoperative computed tomography; E. Processing a radiation therapy treatment plan.

In a preferred embodiment of the invention, a first three-dimensional scan of the region involved in the operation is optionally acquired before surgery (step B‘). Said first scan can advantageously be acquired directly with the patient on the operating bed and does not require cumbersome instrumentation or long acquisition times. By way of example, acquiring a three-dimensional scan takes less than 10 seconds. Said preferred embodiment is diagrammatically depicted in Figure 3. The steps enclosed by the box can be performed during surgery, possibly in the same operating room. Step B‘ and the consequent obtaining of the three-dimensional scan are optional steps.

Step A

Clinical practice dictates that preoperative computed tomography is acquired approximately ten days or a week earlier than the surgery date. However, in principle it is possible to acquire the image even in the hours immediately before the operation since the processing times of the image are around 10s.

Step B.

The three-dimensional scans (B, B‘) can be acquired by any instrument capable of carrying out three-dimensional scans with a spatial resolution of around at least one millimeter over an acquisition field of around least 30x30cm. Preferably, instruments capable of working even under difficult lighting conditions and/or in the presence of reflections are used. These features are met, for example, by the sensors used for facial recognition in modern smartphones.

Step C.

For the purpose of aligning the images, a software algorithm is used, which is capable of mapping the two images on a homogeneous space, also identifying the transformation to be applied to the three- dimensional scan to maximize the alignment between this and the preoperative computed tomography.

As for the correction of the preoperative CT, a software algorithm is used, which takes the two images as an input, the preoperative CT and the three-dimensional scan, respectively, and is capable of “erasing” the pixels of the first to which the pixels of the post-surgery three- dimensional scan correspond, in which there is no more tissue or, on the contrary, to insert new pixels where required.

Examples of these algorithms are present in Python’s SimplelTK libraries (ref: https://simpleitk.org). Step D.

The simulation step (D) is a common practice in the physics of elementary particles and also in the applications thereof in the medical field, and consists in simulating the passage of the radiation of matter, the energy loss thereof, the generation of secondary particles, and all the processes which contribute to the dose release in the various organs.

In a preferred embodiment of the invention, the simulation algorithm consists of Monte Carlo software algorithms or simulation software based on the Monte Carlo method. Examples of these software are FLUKA and Geant4, commonly used in particle physics. Other possible examples of simulation algorithms are the Pencil Beam and Hybrid Monte Carlo algorithms.

Step E.

Radiotherapy treatment means any therapeutic treatment with ionizing radiation, preferably electron beams of energy between 1-13 MeV. Once the best configuration of the treatment to be carried out has been chosen, the TPS is also capable of providing a precise “dosimetry report” related thereto, i.e., an account of how much dose has been transferred to each organ. In fact, this is a key medical document in order to plan any new treatments which the same patient may need later.

Validation of the method

In order to demonstrate the effectiveness of the suggested method, the computed method was tested by simulating a surgical treatment (or surgery) on a breast phantom. A surface image (first three-dimensional scan) of the phantom was acquired.

The surgery was then simulated by removing a portion of the material forming the phantom itself. A second three-dimensional scan was then acquired on the phantom already subjected to the modifications (surgery). The ability to perform a three-dimensional scan with sufficient resolution and field of view for an anatomy “similar” to that expected was achieved by using the facial recognition sensor of a smartphone to acquire the profile of a silicone breast phantom. Both the resolution, and more generally, the logistics of the acquisition were verified to be compatible with the requirements of the technique. As for the image alignment step (d), software has been developed which is capable of taking the two images obtained by 3D scanning of the same phantom, align them, and use the second one to “correct” the first one according to the same criteria reported in step (f); however, in the absence of a true preoperative CT of the phantom, the 3D scan of the intact phantom was used instead of CT. The alignment of the images was obtained by identifying the transformation which maximized the overlap, then applied to the post-surgery 3D scan. The result obtained is a three-dimensional image of the post-surgery phantom which approximates an intraoperative CT.

Application of the practical case to IORT

The computed method was then applied to the practical case in order to process treatment plans for hypothetical IORT operations, thus including radiotherapy. Regarding the Monte Carlo simulation step (D), a Monte Carlo simulation in FLUKA was developed for a treatment with direct electrons at the surgical cavity. To carry out such a simulation, the “corrected” image of the phantom was forcibly superimposed on a segmented CT of a patient, required to evaluate the dosing to the various organs.

Validation of the result

Three Monte Carlo simulations were performed, corresponding to three different values for the density of the pixels created in the blending process. In particular, three “extreme” density values were chosen (min=-200 H.U., mean=-100 H.U., max=0 HU, estimated as extremes of the typical variability of the breast tissue of the patient in hand), to evaluate the maximum intrinsic uncertainty of the technique in hand. Maximum errors of 1% on organ dosing were estimated. Therefore, the intrinsic error of the technique appears to be much lower than the errors currently present in the entire IORT system (~10%).

Advantages With respect to the approach based on true intraoperative imaging

(i.e., CT acquisition between surgery and radiotherapy), that of interest in this patent application has the advantage of a considerably greater simplicity of application. In fact, no classic imaging procedure is necessary, thus saving time, resources, and radiation dosing to the patient. The 3D scanning procedure can be performed in a few seconds, directly in the operating room, requiring no patient movement and without the need for additional expensive equipment (such as “cone beam CT”) beyond the scanning device. Said scanning device can be advantageously integrated into the applicator of the IORT machine. With respect to the approaches of the prior art based on an “a posteriori” modification of the preoperative image, the proposal under discussion has the advantage of a much more realistic rendering of the modified anatomy, which would be exact in volumetric terms, limiting the approximation only to the density of a small part of the tissue moved during the operation. Therefore, the resulting Monte Carlo simulation and the outcomes thereof in terms of dose map would have much greater accuracy and precision.