Domain transformation using semi-supervised CycleGAN for improving performance of classifying thyroid tissue images.

PURPOSE: A large number of research has been conducted on the classification of medical images using deep learning. The thyroid tissue images can be also classified by cancer types. Deep learning requires a large amount of data, but every medical institution cannot collect sufficient number of data for deep learning. In that case, we can consider a case where a classifier trained at a certain medical institution that has a sufficient number of data is reused at other institutions. However, when using data from multiple institutions, it is necessary to unify the feature distribution because the feature of the data differs due to differences in data acquisition conditions. METHODS: To unify the feature distribution, the data from Institution T are transformed to have the closer distribution to that from Institution S by applying a domain transformation using semi-supervised CycleGAN. The proposed method enhances CycleGAN considering the feature distribution of classes for making appropriate domain transformation for classification. In addition, to address the problem of imbalanced data with different numbers of data for each cancer type, several methods dealing with imbalanced data are applied to semi-supervised CycleGAN. RESULTS: The experimental results showed that the classification performance was enhanced when the dataset from Institution S was used as training data and the testing dataset from Institution T was classified after applying domain transformation. In addition, focal loss contributed to improving the mean F1 score the best as a method that addresses the class imbalance. CONCLUSION: The proposed method achieved the domain transformation of thyroid tissue images between two domains, where it retained the important features related to the classes across domains and showed the best F1 score with significant differences compared with other methods. In addition, the proposed method was further enhanced by addressing the class imbalance of the dataset.

Improved artificial intelligence discrimination of minor histological populations by supplementing with color-adjusted images.

Despite the dedicated research of artificial intelligence (AI) for pathological images, the construction of AI applicable to histopathological tissue subtypes, is limited by insufficient dataset collection owing to disease infrequency. Here, we present a solution involving the addition of supplemental tissue array (TA) images that are adjusted to the tonality of the main data using a cycle-consistent generative adversarial network (CycleGAN) to the training data for rare tissue types. F1 scores of rare tissue types that constitute < 1.2% of the training data were significantly increased by improving recall values after adding color-adjusted TA images constituting < 0.65% of total training patches. The detector also enabled the equivalent discrimination of clinical images from two distinct hospitals and the capability was more increased following color-correction of test data before AI identification (F1 score from 45.2 ± 27.1 to 77.1 ± 10.3, p < 0.01). These methods also classified intraoperative frozen sections, while excessive supplementation paradoxically decreased F1 scores. These results identify strategies for building an AI that preserves the imbalance between training data with large differences in actual disease frequencies, which is important for constructing AI for practical histopathological classification.

Follicular cholangitis mimicking a common bile duct cancer: a case report.

BACKGROUND: Follicular cholangitis (FC) is a benign bile duct disease that was first reported 2003. Pathologically, it is characterized by lymphoplasmacytic infiltration with multiple lymphoid follicle formations under the mucosal layer of the biliary tract. However, as this disease is extremely rare, little is known about its etiology and pathogenesis. CASE PRESENTATION: A 77-year-old woman was diagnosed with middle bile duct stenosis and potential increases in alkaline phosphatase (ALP) and γ-glutamyl transpeptidase levels (γ-GTP). Carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9) and IgG4 levels were all within the normal limits. Contrast-enhanced computed tomography (CE-CT) and magnetic resonance imaging (MRI) revealed bile duct dilation from intrahepatic to upper common bile duct and an irregular mass lesion in distal bile duct. Additionally, multiple overlapping leaf-like folds were detected. 18F-fluorodeoxyglucose positron emission tomography-computed tomography (18F-FDG-PET/CT) did not demonstrate fluorodeoxyglucose uptake. Subtotal stomach-preserving pancreaticoduodenectomy with regional lymph node dissection was performed because common bile duct cancer could not be ruled out. The resected specimen showed diffuse homogeneous middle bile duct wall thickening. Microscopically, the lesion exhibited thick fibrosis with several invaded lymphoplasmacytic cells, and lymphoid follicle formations were detected under the mucosal layer. Immunohistochemical staining (IHC) revealed positive for CD3, CD4, CD20 and CD79a, and these findings led to a final diagnosis of FC. The patient has not experienced recurrence to date (42 months postoperatively). CONCLUSIONS: Currently, accurate preoperative diagnosis of FC is difficult. More cases must be accumulated to generate additional knowledge on its precise diagnosis and proper treatment.

Exploration of sepsis assisting parameters in hospital autopsied-patients: a prospective study.

Although Sepsis-3 doesn't require evidence of bacteremia to diagnose sepsis, clinicians often want to identify the causative pathogen at autopsy. In principle, if the blood cultures are the same at ante- and postmortem, the cause of death is obvious. However, interpretations of postmortem blood cultures are often difficult due to discordance, negativity, mixed infection, and contamination, of pathogens occupying ≥ 50% of the tests. To increase specificity identifying agonal phase sepsis in the situations where blood cultures are discordant, multiple or negative at postmortem, we established a scoring system using blood cultures, procalcitonin (PCN) showing highest sensitivity and specificity for postmortem serum, and bone marrow polyhemophagocytosis (PHP). Histological sepsis showed significantly higher levels of culture score (2.3 ± 1.5 vs. 0.4 ± 0.5, p < 0.001), PHP score (2.5 ± 0.8 vs. 1.0 ± 1.1, p < 0.001), and PCN score (1.8 ± 0.8 vs. 0.8 ± 0.6, p < 0.01) than non-septic patients. Receiver operating characteristic curve analysis indicated that estimation of three scores was the most reliable indicator for recognizing agonal phase sepsis. These findings suggest that the combination of these three inspections enables to determine the pathological diagnoses of sepsis even it is not obvious by discordant, mixed or negative blood cultures.

A case of eosinophilic fasciitis without skin manifestations: a case report in a patient with lupus and literature review.

Eosinophilic fasciitis (EF) is a rare connective tissue disease that causes inflammation and fibrosis of the fascia, inducing pain and motor dysfunction. Characteristic skin manifestations, such as edema, erythema, induration, peau d'orange appearance, and the groove sign, are of diagnostic significance and observed in the majority of patients with EF. We herein report a case of EF without these characteristic skin manifestations. A 66-year-old Japanese woman developed progressive limb pain and motor dysfunction. No skin changes were observed. We diagnosed the patient with EF based on the clinical course, magnetic resonance imaging, and en bloc biopsy containing fascia and muscle. Oral prednisolone therapy markedly attenuated limb pain and motor dysfunctions. Through a systemic search of the medical literature, we retrieved 4 juvenile cases and 8 adult cases of EF without characteristic skin manifestations during the clinical course. We herein present a systemic review on EF without skin manifestations and discuss differences between the two proposed sets of diagnostic criteria of EF.

A typical case of resected pancreatic hamartoma: a case report and literature review on imaging and pathology.

BACKGROUND: Pancreatic hamartomas are rare entities and difficult to diagnose before resection. We report a case of resected pancreatic hamartoma and literature review of typical characteristics of the lesion. CASE PRESENTATION: A 78-year-old man presented with a mass in his pancreas, which was incidentally identified when he experienced pneumonia. No remarkable symptoms were observed, and laboratory tests showed no abnormalities, except a slight carcinoembryonic antigen elevation. Enhanced computed tomography and magnetic resonance imaging showed a well-demarcated solid mass with heterogeneous contrast that was 2 cm in size. A gradual enhancement pattern was also observed. The biopsy revealed no specific findings; therefore, surgical resection was necessitated to confirm the diagnosis. Histopathologically, ducts, acinar cells, and adipose cells without atypia were observed among abundant fibrous stroma, but islets of Langerhans and peripheral nerves were absent. An immunohistochemical examination demonstrated CD34 and c-kit positive staining in the stromal cells, S-100 positivity in the adipose cells, and a lack of elastic fibers in the duct walls. The lesion was diagnosed as a pancreatic hamartoma. CONCLUSION: Asymptomatic pancreatic hamartomas can avoid resection. A careful consideration of imaging and appropriate immunohistochemistry of biopsy specimen may facilitate accurate diagnosis before resection. Therefore, sufficient recognition of the characteristics of pancreatic hamartomas is desirable.