Effects of Image Quantity and Image Source Variation on Machine Learning Histology Differential Diagnosis Models.

AIMS: Histology, the microscopic study of normal tissues, is a crucial element of most medical curricula. Learning tools focused on histology are very important to learners who seek diagnostic competency within this important diagnostic arena. Recent developments in machine learning (ML) suggest that certain ML tools may be able to benefit this histology learning platform. Here, we aim to explore how one such tool based on a convolutional neural network, can be used to build a generalizable multi-classification model capable of classifying microscopic images of human tissue samples with the ultimate goal of providing a differential diagnosis (a list of look-alikes) for each entity. METHODS: We obtained three institutional training datasets and one generalizability test dataset, each containing images of histologic tissues in 38 categories. Models were trained on data from single institutions, low quantity combinations of multiple institutions, and high quantity combinations of multiple institutions. Models were tested against withheld validation data, external institutional data, and generalizability test images obtained from Google image search. Performance was measured with macro and micro accuracy, sensitivity, specificity, and f1-score. RESULTS: In this study, we were able to show that such a model's generalizability is dependent on both the training data source variety and the total number of training images used. Models which were trained on 760 images from only a single institution performed well on withheld internal data but poorly on external data (lower generalizability). Increasing data source diversity improved generalizability, even when decreasing data quantity: models trained on 684 images, but from three sources improved generalization accuracy between 4.05% and 18.59%. Maintaining this diversity and increasing the quantity of training images to 2280 further improved generalization accuracy between 16.51% and 32.79%. CONCLUSIONS: This pilot study highlights the significance of data diversity within such studies. As expected, optimal models are those that incorporate both diversity and quantity into their platforms.s.

Bone Strength/Bone Mass Discrepancy in Glucocorticoid-Treated Adult Mice.

Glucocorticoids increase bone fragility in patients in a manner that is underestimated by bone mass measurement. This study aimed to determine if the adult mouse could model this bone strength/bone mass discrepancy. Forty-two 13-week-old BALB/cJ mice were randomized into vehicle and glucocorticoid groups, implanted with vehicle or 6-methylprednisolone pellets, and necropsied after 60 and 120 days. Bone strength and bone mass/microarchitecture were assessed at the right central femur (CF; cortical-bone-rich) and sixth lumbar vertebral body (LVB6; trabecular-bone-rich). Bound water (BW) of the whole right femur was analyzed by proton-nuclear magnetic resonance (1H-NMR) relaxometry. Data were analyzed by two-factor ANOVA with time (day 60 and day 120) and treatment (vehicle and glucocorticoid) as main effects for all data. Significant interactions were further analyzed with a Tukey's post hoc test. Most bone strength measures in the CF were lower in the glucocorticoid group, regardless of the duration of treatment, with no time × treatment interaction. However, bone mass measures in the CF showed a significant time × treatment interaction (p = 0.0001). Bone strength measures in LVB6 showed a time × treatment interaction (p < 0.02) such that LVB6 strength was lower after 120 days of glucocorticoids compared with 120 days of vehicle treatment. Whole-femur-BW was lower with both glucocorticoid treatment (p = 0.0001) and time (p < 0.02), with a significant time × treatment interaction (p = 0.005). Glucocorticoid treatment of male BALB/cJ mice resulted in the lowering of bone strength in both cortical and trabecular bone that either appeared earlier or was greater than the treatment-related changes in bone mass/microarchitecture. The adult mouse may be a good model for investigating the bone strength/mass discrepancy observed in glucocorticoid-treated patients. © 2020 The Authors. JBMR Plus published by Wiley Periodicals LLC. on behalf of American Society for Bone and Mineral Research.

Statins, Fibrates, and Other Peroxisome Proliferator-Activated Receptor Agonists for the Treatment of Cholestatic Liver Diseases.

Primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC) are autoimmune cholestatic liver diseases that commonly result in the need for liver transplantation. The lack of an effective therapy for PSC remains a largely unmet need in hepatology, and although the majority of patients with PBC will have an adequate response to ursodeoxycholic acid and/or obeticholic acid, there is a need for treatment among patients who do not respond completely to these therapies. Investigations of statins, fibrates, and other peroxisome proliferator-activated receptor (PPAR) agonists suggest clinical benefit with some of these agents. Statins have recently been suggested to improve outcomes in patients with PSC but have not demonstrated benefit in patients with PBC, whereas fibrates and newer PPAR agonists appear to improve biochemical markers linked to better clinical outcomes in patients with PBC. Further research is needed to fully understand the clinical efficacy of these agents in the treatment of PBC and PSC.

Prevalence of glucocorticoid induced osteonecrosis in the mouse is not affected by treatments that maintain bone vascularity.

OBJECTIVE: Determine if LLP2A-Ale or PTH (1-34) affects the prevalence of glucocorticoid-induced osteonecrosis (ON) in a mouse model. METHODS: Eight-week-old young adult male BALB/cJ mice were weight-randomized into Control (Con), glucocorticoid (GC)-only, or concurrent treatments with GC and LLP2A-Ale (250 µg/kg or 500 µg/kg, IV, Days 1, 14, 28) or parathyroid hormone hPTH (1-34) (40 µg/kg, 5×/week). Mice were necropsied after 45 days for qualitative evaluation of prevalent ON and quantitative evaluation of vascularity in the distal femoral epiphysis (DFE); and quantitative evaluation of bone mass, microarchitecture, and strength in the distal femoral metaphysis and lumbar vertebral body. RESULTS: The prevalence of ON was 14% in the Con group and 36% in the GC-only group (P = 0.07). The prevalence of ON did not differ among GC-only, GC + LLP2A-Ale, and GC + PTH groups. GC-only mice had significantly lower trabecular and cortical bone strength than Con, while GC + LLP2A-Ale (500 µg/kg) and GC + PTH (1-34) groups had significantly greater trabecular bone strength than the GC-only group. GC + LLP2A-Ale (250 µg/kg and 500 µg/kg) and GC + PTH had significantly higher trabecular bone volume than GC-only mice at the vertebrae, distal femoral epiphyses and distal femoral metaphyses. DFE vascularity was lower in GC-only mice than in all other groups. CONCLUSION: Neither LLP2A-Ale nor hPTH (1-34) reduced the prevalence of GC-induced ON, compared to GC-only mice. However, GC-treated mice given LLP2A-Ale or hPTH (1-34) had better bone mass, microarchitecture, and strength in trabecular-rich regions, and higher levels of vascularity than GC-only mice.

Effect of the Specific Carbohydrate Diet on the Microbiome of a Primary Sclerosing Cholangitis and Ulcerative Colitis Patient.

A 20-year-old female was diagnosed with ulcerative colitis (UC) at age 14 and primary sclerosing cholangitis (PSC) at age 16. The PSC was successfully treated with high doses of oral vancomycin; however, the UC was more difficult to manage. After many drug treatments failed to treat the UC, the patient began following the specific carbohydrate diet (SCD). This report documents fecal microbiome changes resulting from following the SCD for two weeks. The DNA extracted from fecal samples was subjected to 16S rRNA gene sequencing to quantify bacterial species abundance. Not only were substantial changes in the fecal bacterial composition detectable within two weeks, but all UC symptoms were also controlled as early as one week following the start of the diet. The patient's fecal microbiota was dramatically different from those of three healthy control subjects and showed remarkable loss of bacterial diversity in terms of species richness, evenness, and overall diversity measures. Other specific changes in bacterial composition included an increase in Enterobacteriaceae, including Escherichia and Enterobacter species. A two- to three-fold decrease was observed in the prevalence of the most dominant fecal bacterial species, Fusobacterium ulcerans, after two weeks on the SCD. Overall species diversity and evenness increased to levels near the controls, although species richness remained low. These findings provide information on the fecal bacteria from a patient with PSC and UC, following prolonged oral vancomycin treatment, and identifies a potentially specific microbial effect for the SCD.

Osteoporosis in Rheumatic Diseases: Anti-rheumatic Drugs and the Skeleton.

Osteoporosis in rheumatic diseases is a very well-known complication. Systemic inflammation results in both generalized and localized bone loss and erosions. Recently, increased knowledge of inflammatory process in rheumatic diseases has resulted in the development of potent inhibitors of the cytokines, the biologic DMARDs. These treatments reduce systemic inflammation and have some effect on the generalized and localized bone loss. Progression of bone erosion was slowed by TNF, IL-6 and IL-1 inhibitors, a JAK inhibitor, a CTLA4 agonist, and rituximab. Effects on bone mineral density varied between the biological DMARDs. Medications that are approved for the treatment of osteoporosis have been evaluated to prevent bone loss in rheumatic disease patients, including denosumab, cathepsin K, bisphosphonates, anti-sclerostin antibodies and parathyroid hormone (hPTH 1-34), and have some efficacy in both the prevention of systemic bone loss and reducing localized bone erosions. This article reviews the effects of biologic DMARDs on bone mass and erosions in patients with rheumatic diseases and trials of anti-osteoporotic medications in animal models and patients with rheumatic diseases.