Performance of a deep neural network in teledermatology: a single-centre prospective diagnostic study.

BACKGROUND: The use of artificial intelligence (AI) algorithms for the diagnosis of skin diseases has shown promise in experimental settings but has not been yet tested in real-life conditions. OBJECTIVE: To assess the diagnostic performance and potential clinical utility of a 174-multiclass AI algorithm in a real-life telemedicine setting. METHODS: Prospective, diagnostic accuracy study including consecutive patients who submitted images for teledermatology evaluation. The treating dermatologist chose a single image to upload to a web application during teleconsultation. A follow-up reader study including nine healthcare providers (3 dermatologists, 3 dermatology residents and 3 general practitioners) was performed. RESULTS: A total of 340 cases from 281 patients met study inclusion criteria. The mean (SD) age of patients was 33.7 (17.5) years; 63% (n = 177) were female. Exposure to the AI algorithm results was considered useful in 11.8% of visits (n = 40) and the teledermatologist correctly modified the real-time diagnosis in 0.6% (n = 2) of cases. The overall top-1 accuracy of the algorithm (41.2%) was lower than that of the dermatologists (60.1%), residents (57.8%) and general practitioners (49.3%) (all comparisons P < 0.05, in the reader study). When the analysis was limited to the diagnoses on which the algorithm had been explicitly trained, the balanced top-1 accuracy of the algorithm (47.6%) was comparable to the dermatologists (49.7%) and residents (47.7%) but superior to the general practitioners (39.7%; P = 0.049). Algorithm performance was associated with patient skin type and image quality. CONCLUSIONS: A 174-disease class AI algorithm appears to be a promising tool in the triage and evaluation of lesions with patient-taken photographs via telemedicine.

Role of tumour location and surgical extent on prognosis in T2 gallbladder cancer: an international multicentre study.

BACKGROUND: In gallbladder cancer, stage T2 is subdivided by tumour location into lesions on the peritoneal side (T2a) or hepatic side (T2b). For tumours on the peritoneal side (T2a), it has been suggested that liver resection may be omitted without compromising the prognosis. However, data to validate this argument are lacking. This study aimed to investigate the prognostic value of tumour location in T2 gallbladder cancer, and to clarify the adequate extent of surgical resection. METHODS: Clinical data from patients who underwent surgery for gallbladder cancer were collected from 14 hospitals in Korea, Japan, Chile and the USA. Survival and risk factor analyses were conducted. RESULTS: Data from 937 patients were available for evaluation. The overall 5-year disease-free survival rate was 70·6 per cent, 74·5 per cent for those with T2a and 65·5 per cent among those with T2b tumours (P = 0·028). Regarding liver resection, extended cholecystectomy was associated with a better 5-year disease-free survival rate than simple cholecystectomy (73·0 versus 61·5 per cent; P = 0·012). The 5-year disease-free survival rate was marginally better for extended than simple cholecystectomy in both T2a (76·5 versus 66·1 per cent; P = 0·094) and T2b (68·2 versus 56·2 per cent; P = 0·084) disease. Five-year disease-free survival rates were similar for extended cholecystectomies including liver wedge resection versus segment IVb/V segmentectomy (74·1 versus 71·5 per cent; P = 0·720). In multivariable analysis, independent risk factors for recurrence were presence of symptoms (hazard ratio (HR) 1·52; P = 0·002), R1 resection (HR 1·96; P = 0·004) and N1/N2 status (N1: HR 3·40, P < 0·001; N2: HR 9·56, P < 0·001). Among recurrences, 70·8 per cent were metastatic. CONCLUSION: Tumour location was not an independent prognostic factor in T2 gallbladder cancer. Extended cholecystectomy was marginally superior to simple cholecystectomy. A radical operation should include liver resection and adequate node dissection.

ANTECEDENTES: En el cáncer de vesícula biliar, la ubicación del tumor subdivide el estadio T2 en tumores con invasión del lado peritoneal y del lado del hígado (T2a y T2b). Para los tumores que invaden el lado peritoneal (T2a) se sugiere que se puede obviar la resección hepática sin que ello comprometa el pronóstico. Sin embargo, este argumento no ha sido validado. El estudio tuvo como objetivo investigar el valor pronóstico de la localización del tumor en el cáncer de vesícula biliar T2 y establecer la extensión adecuada de la resección quirúrgica. MÉTODOS: Se recogieron los datos clínicos de pacientes que se sometieron a cirugía por cáncer de vesícula biliar en 14 hospitales de Corea, Japón, Chile y Estados Unidos. Se realizaron análisis de la supervivencia y de los factores de riesgo. RESULTADOS: Se dispuso de datos de 937 pacientes para ser evaluados. La tasa de supervivencia global libre de enfermedad a los 5 años fue del 70,6%, y las de T2a y T2b del 74,5% y 65,5% (P = 0,028). Con respecto a la resección hepática, la colecistectomía extendida presentó una tasa mejor de supervivencia libre de enfermedad a los 5 años que la colecistectomía simple (73,0% versus 61,5%, P = 0,012). La tasa de supervivencia libre de enfermedad a los 5 años fue marginalmente mejor para la colecistectomía extendida que para la colecistectomía simple tanto en T2a (76,5% versus 66,1%, P = 0,094) como en T2b (68,2% versus 56,2%, P = 0,084). Las tasas de supervivencia libre de enfermedad a los 5 años no fueron diferentes entre la resección hepática en cuña y la segmentectomía S4b+S5 (74,1% versus 71,5%, P = 0,720). En el análisis multivariable, los factores de riesgo independientes para la recidiva fueron la presencia de síntomas (cociente de riesgos instantáneos, hazard ratio, HR 1,52, P = 0,002), la resección R1 (HR 1,96, P = 0,004) y el estadio N1/N2 (N1 HR 3,40, P < 0,001; N2 HR 9,56, P < 0,001). El 70,8% de las recidivas eran metastásicas. CONCLUSIÓN: La localización del tumor no fue un factor pronóstico independiente en el cáncer de vesícula biliar T2. La colecistectomía extendida fue marginalmente superior que la colecistectomía simple. La cirugía radical debe incluir una resección hepática y una linfadenectomía adecuada.

Effects of brazilin on GLUT4 recruitment in isolated rat epididymal adipocytes.

The effects of brazilin on glucose transport into isolated rat epididymal adipocytes were investigated. Brazilin increased [3H]2-deoxy-D-glucose uptake, which was characterized by an increase in Vmax with no effect on the Km value. Phenylarsine oxide, which inhibits the translocation of glucose transporters, decreased brazilin-stimulated glucose transport to the basal level. The inhibition of phosphatidylinositol 3-kinase (PI3-kinase) with wortmannin also blocked brazilin-stimulated glucose transport. Western blot analysis with an anti-GLUT4 antibody revealed that brazilin increased the translocation of GLUT4 from intracellular pools to the plasma membrane. Brazilin, in combination with phorbol ester, showed an additive effect on glucose transport. The stimulating effect of phorbol ester on glucose transport was inhibited by staurosporine, but the effect of brazilin remained unchanged. Protein kinase C activity was not influenced by brazilin treatment. The inhibition of protein synthesis showed no effect on brazilin-stimulated glucose transport, and GLUT4 content in the total membrane fraction was not altered as a result of treatment with brazilin for 4 hr. Metabolic labeling of GLUT4 with [35S]methionine showed that de novo synthesis of GLUT4 was not induced by brazilin. These data suggest that brazilin may increase glucose transport by recruitment of GLUT4 from intracellular pools to the plasma membrane of adipocytes via the activation of PI3-kinase. However, the effect of brazilin may not be mediated by GLUT4 synthesis and protein kinase C activation.