Improving the performance of convolutional neural network for skin image classification using the response of image analysis filters

Review Research of Medical Image Analysis Using Deep Learning

“…It uses to diagnose of gastrointestinal diseases through a sensor which is quite small to swallow and capture every scenes of anatomical parts that pass through them [41]. Dermoscopic image is another useful modality and is dermoscopic images that use to skin lesion [42], [43]. Breast cancer (BrC) image is another type of well-known cancers that rely on such medical image modalities as mammography which known as X-ray of breast, US which is called sonogram [44].…”

“…For super-pixel image analysis, different structure detection required. This engages image augmentation to aid CNN to extract the features from the original dermoscopy image data [43].…”

Μέθοδοι Υπολογιστικής Νοημοσύνης Για Εφαρμογές Σε Περιβάλλοντα Διάχυτης Νοημοσύνης

“…Τα ΤΝΔΣ είναι μια μέθοδος της Μηχανικής Μάθησης που χρησιμοποιείται σε πολλούς και διαφορετικούς ερευνητικούς τομείς (ταξινόμηση εικόνων [80], αναγνώριση αντικειμένων [52], αναγνώριση ανωμαλιών σε ιατρικές εικόνες [46], αναγνώριση κειμένων [50], κ.α.) βρίσκοντας εφαρμογή και στη βιομηχανία, με σκοπό την επίλυση προβλημάτων του πραγματικού κόσμου με την βοήθεια έξυπνων μηχανών [42,81].…”

Algorithms for automated image and video analysis with applications

“…Also, on this issue, in , we investigate the value of augmenting CNN inputs with the response of mid-level computer vision filters, against the baseline RGB pixel values input, for the detection of malignant melanomas against nevus skin lesions in dermoscopic images. Furthermore, in (Georgakopoulos, 2019), we augment the input data to CNN, in the context of dermoscopy feature classification of skin lesions, for the binary classification into "Malignant" and "Non-malignant" (nevus skin lesions) cases. In 115 addition, we experiment with the classification of superpixels in 4 classes representing the differential structures that appear in skin lesions, namely streaks, pigment network, Milia cysts, and negative network.…”

“…An extra issue we address in this work, involves the detection of the four (4) differential structures in the context of dermoscopy feature classification at a superpixel level, as input data to CNN (Georgakopoulos, 2019). Consequently, the available dataset is quite extensive and no data augmentation by geometric, or other transformations, or the TL is necessary.…”