Urine cell image recognition using a deep-learning model for an automated slide evaluation system.

OBJECTIVES: To develop a classification system for urine cytology with artificial intelligence (AI) using a convolutional neural network algorithm that classifies urine cell images as negative (benign) or positive (atypical or malignant). PATIENTS AND METHODS: We collected 195 urine cytology slides from consecutive patients with a histologically confirmed diagnosis of urothelial cancer (between January 2016 and December 2017). Two certified cytotechnologists independently evaluated and labelled each slide; 4637 cell images with concordant diagnoses were selected, including 3128 benign cells (negative), 398 atypical cells, and 1111 cells that were malignant or suspicious for malignancy (positive). This pathologically confirmed labelled dataset was used to represent the ground truth for AI training/validation/testing. Customized CutMix (CircleCut) and Refined Data Augmentation were used for image processing. The model architecture included EfficientNet B6 and Arcface. We used 80% of the data for training and validation (4:1 ratio) and 20% for testing. Model performance was evaluated with fivefold cross-validation. A receiver-operating characteristic (ROC) analysis was used to evaluate the binary classification model. Bayesian posterior probabilities for the AI performance measure (Y) and cytotechnologist performance measure (X) were compared. RESULTS: The area under the ROC curve was 0.99 (95% confidence interval [CI] 0.98-0.99), the highest accuracy was 95% (95% CI 94-97), sensitivity was 97% (95% CI 95-99), and specificity was 95% (95% CI 93-97). The accuracy of AI surpassed the highest level of cytotechnologists for the binary classification [Pr(Y > X) = 0.95]. AI achieved >90% accuracy for all cell subtypes. In the subgroup analysis based on the clinicopathological characteristics of patients who provided the test cells, the accuracy of AI ranged between 89% and 97%. CONCLUSION: Our novel AI classification system for urine cytology successfully classified all cell subtypes with an accuracy of higher than 90%, and achieved diagnostic accuracy of malignancy superior to the highest level achieved by cytotechnologists.

A Fully Automated Artificial Intelligence System to Assist Pathologists' Diagnosis to Predict Histologically High-grade Urothelial Carcinoma from Digitized Urine Cytology Slides Using Deep Learning.

BACKGROUND: Urine cytology, although a useful screening method for urothelial carcinoma, lacks sensitivity. As an emerging technology, artificial intelligence (AI) improved image analysis accuracy significantly. OBJECTIVE: To develop a fully automated AI system to assist pathologists in the histological prediction of high-grade urothelial carcinoma (HGUC) from digitized urine cytology slides. DESIGN, SETTING, AND PARTICIPANTS: We digitized 535 consecutive urine cytology slides for AI use. Among these slides, 181 were used for AI development, 39 were used as AI test data to identify HGUC by cell-level classification, and 315 were used as AI test data for slide-level classification. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Out of the 315 slides, 171 were collected immediately prior to bladder biopsy or transurethral resection of bladder tumor, and then outcomes were compared with the histological presence of HGUC in the surgical specimen. The primary aim was to compare AI prediction of the histological presence of HGUC with the pathologist's histological diagnosis of HGUC. Secondary aims were to compare the time required for AI evaluation and concordance between the AI's classification and pathologist's cytology diagnosis. RESULTS AND LIMITATIONS: The AI capability for predicting the histological presence of HGUC was 0.78 for the area under the curve. Comparing the AI predictive performance with pathologists' diagnosis, the AI sensitivity of 63% for histological HGUC prediction was superior to a pathologists' cytology sensitivity of 46% (p = 0.0037). On the contrary, there was no significant difference between the AI specificity of 83% and pathologists' specificity of 89% (p = 0.13), and AI accuracy of 74% and pathologists' accuracy of 68% (p = 0.08). The time required for AI evaluation was 139 s. With respect to the concordance between the AI prediction and pathologist's cytology diagnosis, the accuracy was 86%. Agreements with positive and negative findings were 92% and 84%, respectively. CONCLUSIONS: We developed a fully automated AI system to assist pathologists' histological diagnosis of HGUC using digitized slides. This AI system showed significantly higher sensitivity than a board-certified cytopathologist and may assist pathologists in making urine cytology diagnoses, reducing their workload. PATIENT SUMMARY: In this study, we present a deep learning-based artificial intelligence (AI) system that classifies urine cytology slides according to the Paris system. An automated AI system was developed and validated with 535 consecutive urine cytology slides. The AI predicted histological high-grade urothelial carcinoma from digitized urine cytology slides with superior sensitivity than pathologists, while maintaining comparable specificity and accuracy.

Effectiveness of the forced-choice coin test for detecting malingering during forensic psychiatric examinations.

Background: In general clinical psychiatric practice, open questions are favored over closed ones because they are considered more therapeutically effective and less likely to make the patients pander to us. However, in forensic psychiatric examinations, suspects may attempt malingering. Case Presentation: Using a simple examination based on a forced-choice technique, the author proved that the level of intelligence of a theft suspect pretending to have an intellectual developmental disorder was not so low. The author prepared two sets consisting of a few coins each and asked the suspect to choose which set had a higher total value. The suspect was questioned repeatedly over multiple trials. He always selected the wrong set over the course of more than 10 trials. Conclusion: If the suspect really did not know the correct answer, the probability of getting the answer right or wrong in a binary choice question is 50% for both. The probability of answering the question wrong by chance 10 times in a row is (1/2),10 in other words, about 0.1%. It was evident that the suspect intentionally kept answering incorrectly. When suspects who pretend to have an intellectual developmental disorder answer only "I do not know" to all questions without actively playing out the symptoms, it is difficult to demonstrate that the symptoms are psychiatrically conflicting and prove that they are malingering. Even in such cases, this type of test based on a forced-choice technique can be used to prove that suspects are behaving falsely.

Novel device for dividing core needle biopsy specimens to provide paired mirror image-like tissues for genetic and pathological tests.

In a world that seeks precision medicine, genetic testing is gaining importance in clinical decision making. We previously reported the utility of a novel tool for longitudinally dividing core needle biopsy (CNB) tissues into two filamentous tissues that can provide paired mirror image-like tissues (mirror-tissues) that spatially match each other. In this study, we investigated its application in gene panel testing in patients who underwent prostate CNB. Four hundred and forty-three biopsy cores were obtained from 40 patients. Of them, 361 biopsy cores (81.5%) were judged by a physician to be appropriate for dividing into two pieces using the new device, of which a histopathological diagnosis was successfully reached in 358 biopsy cores (99.2%). Among them, the quality and quantity of nucleic acid in 16 appropriately divided cores were assessed and found to be sufficient for gene panel testing, and histopathological diagnosis was successfully obtained from the remaining divided cores. The novel device for longitudinally-dividing CNB tissue provided mirror image-like paired-tissues for gene panel and pathology testing. The device might be a promising tool for obtaining genetic and molecular biological information, in addition to histopathological diagnosis, helping to advance personalized medicine.

Increased Incidence of Inflammatory Bowel Disease in Association with Dietary Transition (Westernization) in Japan.

INTRODUCTION: Inflammatory bowel disease has become a global disease, but its key environmental factors still remain unrecognized. This study aimed to clarify the role of dietary transition (westernization) in the increased incidence of inflammatory bowel disease in Japan. METHODS: Annual numbers of new cases of inflammatory bowel disease in Japan over the period from 1965 to 2000 found in a nationwide database compiled by the government and the daily amount of food and nutrient intake per capita for the same period revealed by the National Nutrition Survey have been used to analyze their interrelation. RESULTS: Rapid increases in the estimated incidence per 100,000 population have been observed, that is, from 0.08 in 1965 to 4.8 in 2000 for ulcerative colitis and from 0.003 to 1.3 in 2000 for Crohn's disease, with an extremely high correlation between the annual numbers of new cases of the respective diseases (r = 0.970). Intake of both animal fat and animal protein increased, while intake of rice decreased during the period. Of all food groups, the intake of rice as a staple food showed the highest negative correlation coefficient with the numbers of new cases of both ulcerative colitis (r = -0.825, 95% CI: -0.908 to -0.681, p < 0.0001) and Crohn's disease (r = -0.836, 95% CI: -0.914 to -0.700, p < 0.0001). CONCLUSIONS: An increased incidence of inflammatory bowel disease was observed to coincide with dietary westernization in Japan. Our results support the assertion that dietary westernization is a key environmental factor in inflammatory bowel disease.

Usefulness of a novel device to divide core needle biopsy specimens in a spatially matched fashion.

We developed a novel dividing device that can split needle biopsy tissues along longitude axis aiming to achieve definitive molecular-biological and genetical analysis with reference of pathological diagnosis of the side-by-side divided tissue as spatially matched information. The aim of this study was to evaluate the feasibility and potential usefulness of the novel dividing device to provide the appropriate materials for molecular diagnosis. The new device was examined using mouse xenograft tumors. Real-time quantitative PCR and genetic test were performed to evaluate the feasibility and usefulness of the device. All the samples from needle biopsy were successfully divided into two pieces. Quality and quantity from divided samples harbor high enough to perform gene expression analysis (real-time PCR) and genetic test. Using two divided samples obtained from xenograft tumor model by needle biopsy, the % length of xenograft tumor (human origin) was significantly correlated with the % human genomic DNA (p = 0.00000608, r = 0.987), indicating that these divided samples were spatially matched. The novel longitudinally dividing device of a needle biopsy tissue was useful to provide the appropriate materials for molecular-biological and genetical analysis with reference of pathological diagnosis as spatially matched information.

Stimulatory effect of Shoyu polysaccharides from soy sauce on the intestinal immune system.

Soy sauce (Shoyu) is a traditional Japanese fermented seasoning and is available worldwide. We investigated the effect of Shoyu polysaccharides (SPS) prepared from soy sauce on the intestinal immune system of mice. SPS enhanced the production of immunoglobulin A (IgA) from Peyer's patch cells in vitro, and its oral administration to 7-week-old male BALB/c mice for 2 weeks at a dose of 1.5 mg per day significantly (p<0.01) increased the concentration of IgA in the intestine as compared to control mice. Furthermore, experiments on the intestinal transport of SPS in vitro using the human intestinal epithelial cell line Caco-2 confirmed the permeation of uronic acid to be time-dependent. In conclusion, SPS of soy sauce enhanced the production of IgA in vitro and in vivo, and the digested SPS might cross the enterocytic monolayer. Thus, soy sauce is a potentially promising food for enhancing host defenses.

Hypolipidemic effect of Shoyu polysaccharides from soy sauce in animals and humans.

Soy sauce (Shoyu) is a traditional Japanese fermented seasoning and is available worldwide. We investigated the effects of Shoyu polysaccharides (SPS) prepared from soy sauce on hyperlipidemia in vitro and in vivo. First, SPS inhibited pancreatic lipase. Second, in experiments with animals, it was found that SPS reduced serum triacylglycerol (TG) elevation induced by high-fat diets. Third, in a 2-week placebo-controlled parallel group study, healthy men (TG <150 mg/dl) were treated with 600 mg of SPS (n=5) or placebo (n=5) every day. After 2 weeks, serum TG elevation was significantly (P<0.05) lower in the SPS-treated group than in the placebo-treated group after 6 h of a high-fat diet. Fourth, in a 4-week randomized, double-blind, placebo-controlled parallel group study, hyperlipidemic men (TG >150 mg/dl) were treated with 600 mg of SPS (n=15) or placebo (n=15) daily. After 4 weeks, serum TG levels in the SPS-treated group were significantly (P<0.05) lower than the baseline (0 week). In conclusion, SPS of soy sauce reduce lipid absorption, and soy sauce is a potentially promising seasoning for the treatment of hyperlipidemia through food.

Promotive effect of Shoyu polysaccharides from soy sauce on iron absorption in animals and humans.

Soy sauce (Shoyu) is a traditional fermented seasoning of Japan and is available throughout the world. We investigated the effect of Shoyu polysaccharides (SPS) prepared from soy sauce on iron absorption in vitro and in vivo. First, by measuring the iron-binding capacity of SPS, it was found that SPS stabilized the solubility of ferrous iron at neutral pH's by forming a complex, Fe-SPS. Second, in experiments with animals, it was found that SPS enhanced the absorption and/or pooling of iron in organs when anemic rats were fed iron-supplemented diets. Third, in a 4-week randomized, double-blind, placebo-controlled parallel group study, healthy women were treated with 600 mg of SPS (n = 22) or placebo (n = 23) each day. After the 4 weeks, serum levels of iron, hematocrit, and hemoglobin were significantly higher (P < 0.05) in the SPS-treated than in the placebo-treated group. In conclusion, SPS of soy sauce enhanced iron absorption, and soy sauce is a potentially promising seasoning for the treatment of anemia through food.

Bactericidal effects of konjac fluid on several food-poisoning bacteria.

In this study, the bactericidal effects of Japanese alkaline foods on food-poisoning bacteria were evaluated. Konjac is an alkaline food soaked in calcinated calcium (the pH of konjac fluid ranges from 11.42 to 12.53). Konjac fluids completely inactivated Escherichia coli, enterohemorrhagic E. coli O157:H7 and E. coil O26:H9, Salmonella Enteritidis, Vibrio parahemolyticus. and Staphylococcus aureus. The initial level of 6 log CFU/ml dramatically decreased after incubation with konjac fluid, and no viable gram-negative bacterium cells could be detected within 1 to 2 days and no viable S. aureus cells could be detected within 3 to 5 days. On the other hand, treatment with konjac fluid was also effective in reducing levels of spore-forming bacteria (Bacillus subtilis, Bacillus cereus, Clostridium perfringens, and Clostridium botulinum type E and type A). At least a 4-log reduction of spore-forming bacteria was obtained in konjac fluid within 7 to 14 days. Vegetative cells were more susceptible to konjac fluid than spores were. When the initial cell count was 6 log CFU/ml, a few surviving spores remained for 60 to 90 days, but no spores could be detected after 120 days. When the initial count of spore-forming bacteria was 3 to 4 log CFU/ml, the cells considered vegetative were completely inactivated within I to 3 days. Repeated treatment with konjac fluid caused complete inactivation of spores in less than 1 to 3 days. Our studies indicate that konjac fluid, which has a long history of use in food, will control food-poisoning bacterial contamination during the production or preservation of konjac and other foods and has a preventive effect on bacteria that can cause severe disease at uniquely low levels.