Tissue Segmentation from Whole-Slide Images Using Lightweight Neural Networks

Pathology slides of malignancies are segmented using lightweight convolutional neural networks (CNNs) that may be deployed on mobile devices. This is made possible by preprocessing candidate images to make CNN analysis tractable and also to exclude regions unlikely to be diagnostically relevant. In a training phase, labeled whole-slide histopathology images are first downsampled and decomposed into square tiles. Tiles corresponding to diseased regions are analyzed to determine boundary values of a visual criterion, image entropy. A lightweight CNN is then trained to distinguish tiles of diseased and non-diseased tissue, and if more than one disease type is present, to discriminate among these as well. A segmentation is generated by downsampling and tiling a candidate image, and retaining only those tiles with values of the visual criterion falling within the previously established extrema. The sifted tiles, which now exclude much of the non-diseased image content, are efficiently and accurately classified by the trained CNN. Tiles classified as diseased tissue ¾ or in the case of multiple possible subtypes, as the dominant subtype in the tile set ¾ are combined, either as a simple union or at a pixel level, to produce a segmentation mask or map. This approach was applied successfully to two very different datasets of large whole-slide images, one (PAIP2020) involving multiple subtypes of colorectal cancer and the other (CAMELYON16) single-type breast-cancer metastases. Scored using standard similarity metrics, the segmentations exhibited notably high recall, even when tiles were large relative to tumor features.

The frequency of demand and options for surgical treatment of breast cancer metastases in long and flat bones, depending on the histological subtype of the tumor

Introduction. Bone metastases increase the risk of developing a pathological fracture, spinal cord compression, pain syndrome and require palliative, radiation therapy and oncoortopedic operations. The appearance of these metastases and the above complications is associated with an unfavorable prognosis and negatively affects the quality of life of patients.Objective – to analyze the outcomes of surgical treatment in patients with breast cancer metastases to long and flat bones and to identify an optimal surgical tactics depending on the morphological subtype of the tumor.Materials and methods. Between April 2015 and April 2021, a total of 731 patients with breast cancer and bone metastases were consulted in the departments of the Russian Research Center of Radiology, Ministry of Health of Russia. Two hundred and thirty patients (31.5 %) had indications for orthopedic surgery. This study included 78 patients with breast cancer metastases to long or flat bones. Of them, 50 individuals had metastases to bones only, whereas 28 participants additionally had metastases to other organs. Luminal A and B breast cancer was diagnosed in 27 patients (34.6 %) and 40 patients (51.2 %), respectively. Ten participants (12.8 %) were found to have HER2-positive tumors, whereas one patient (1.3 %) had triplenegative cancer.Results. Eighteen patients with metastatic lesions in long bones have undergone bone grafting; 17 patients had intramedullary internal fixation with osteoplasty; and 1 patient had extramedullary internal fixation with osteoplasty. Forty two patients with metastatic lesions in the flat bones have undergone 39 osteoplasty surgeries of the iliac bone, while 3 patients had osteoplasty surgeries of the sternum. The survival rates of patients with luminal A, luminal B, HER2- positive, and triple-negative breast cancer were as follows: 1-year survival – 95, 84, 87 and 100 %, respectively; 3-year survival – 83, 64, 23 and 0 %, respectively; 5-year survival – 66, 32, 23 and 0 %, respectively. Thirty-one patients were lost to follow-up between month 0 and month 46. Eighteen patients died between month 1 and month 58.Conclusion. Patients with luminal A and B breast cancer and threatened pathologic fracture should undergo intramedullary/extramedullary internal fixation, while for patients who already developed pathologic fracture and their estimated life expectancy does not exceed 1 months, it is recommended to perform bone grafting regardless of the tumor immunohistochemical type. Patients with threatened pathologic fractures demonstrated a more favorable prognosis, better functional status, and higher survival rates than patients with pathologic fractures: 1-year survival was 90 and 87 %, respectively; 3-year survival was 67 and 58 %, respectively, and 5-year survival was 54 and 0 %, respectively. Radiation therapy increases the risk of pathological fractures, but is not a contraindication for orthopedic surgery.