CLAIMS

1 . A computer-implemented method capable of automatically detecting and differentiating esophageal , stomach , small bowel and colon lesions in device-assisted enteroscopy images by classifying the pixels as esophageal , stomach , small bowel and colon lesions , comprising selecting the architecture combination and fully training such architecture for esophageal , stomach , small bowel and colon lesions with means of output validation and storage capabilities wherein the method : selects a number of subsets of all endoscopic ultrasonography images /videos , each of said subsets considering only images from the same patient ;

- selects another subset as validation set , wherein the subset does not overlap chosen images on the previously selected subsets ;

- pre-trains (8000) of each of the chosen subsets with one of a plurality of combinations of image feature extraction component , followed by a subsequent classification neural network component for pixel classification as esophageal , stomach , small bowel and colon lesions wherein said pre- training ;

- early stops when the scores do not improved over a given number of epochs , namely three ; evaluates the performance of each of the combinations ;

- is repeated on new, different subsets , with another networks combination and training hyperparameters , wherein such new combination considers a higher number of dense layers if the fl-metric is low

AMENDED SHEET (ARTICLE 19) and fewer dense layers if the fl -metric suggests overfitting ; selects (400) the architecture combination that performs best during pre- training ; fully trains and validates ( 9000) the selected architecture combination using the entire set of deviceassisted enteroscopy images to obtain an optimized architecture combination for each classification ;

- predicts ( 6000) esophageal , stomach , small bowel and colon lesions using said optimized architecture combination for classification ; receives the classification output (270) of the prediction module ( 6000) by an output collect module (7000) with means of communication to a third-party capable of performing validation of said network classification ;

- stores the corrected prediction into the storage component (2000) .

2 . The method of claim 1 , wherein the pre-training (8000) is performed dividing and selecting the entire device-assisted enteroscopy images cohort into a predefined amount of folds , wherein said given patient specific images are contained in one and one only fold .

3. The method of claim 1 , wherein the classification network architecture comprises at least two blocks , each having a Dense layer followed by a Dropout layer .

4 . The method of claims 1 and 3 , wherein the last block of the classification component includes a BatchNormalization layer followed by a Dense layer where the depth size is equal to the number of lesions type one desires to classify .

5 . The method of claim 1 , wherein the set of pre-trained neuronal networks is the best performing among the following :

22

AMENDED SHEET (ARTICLE 19) VGG16 , InceptionVB , Xception , EfficientNetB5 ,

EfficientNetB7 , Resnet50 and Resnetl25 .

6. The method of claims 1 and 5 , wherein the best performing combination is chosen based on the overall accuracy and on the fl-metrics .

7 . The method of claims 1 and 5 , wherein the training of the best performing combination comprises two to four dense layers in sequence , starting with 4096 and decreasing in half up to 512 .

8 . The method of claims 1 , 5 and 7 , wherein between the final two layers of the best performing combination there is a dropout layer of 0 . 1 drop rate .

9 . The method of claim 1 , wherein the training of the samples includes a ratio of training-to-validation of 10%-90% .

10 . The method of claim 1 , wherein the validation of the network output is achieved by interpreting the accuracy of the classification output and of correcting a wrong prediction , wherein the third-party comprises at least one of : another neuronal network , any other computational system adapted to perform the validation task or , optionally, a physician expert in gastroenterological imagery;

11 . The method of claim 10 , wherein the third-party validation is done by user-input .

12 . The method of claim 1 wherein the training dataset includes images in the storage component (2000) that were predicted sequentially performing the steps of such method .

13. A portable endoscopic device comprising instructions which , when executed by a processor , cause the computer to carry out the steps of the method of claims 1-12 .

23

AMENDED SHEET (ARTICLE 19)