Man against machine reloaded: performance of a market-approved convolutional neural network in classifying a broad spectrum of skin lesions in comparison with 96 dermatologists working under less artificial conditions.

BACKGROUND: Convolutional neural networks (CNNs) efficiently differentiate skin lesions by image analysis. Studies comparing a market-approved CNN in a broad range of diagnoses to dermatologists working under less artificial conditions are lacking. MATERIALS AND METHODS: One hundred cases of pigmented/non-pigmented skin cancers and benign lesions were used for a two-level reader study in 96 dermatologists (level I: dermoscopy only; level II: clinical close-up images, dermoscopy, and textual information). Additionally, dermoscopic images were classified by a CNN approved for the European market as a medical device (Moleanalyzer Pro, FotoFinder Systems, Bad Birnbach, Germany). Primary endpoints were the sensitivity and specificity of the CNN's dichotomous classification in comparison with the dermatologists' management decisions. Secondary endpoints included the dermatologists' diagnostic decisions, their performance according to their level of experience, and the CNN's area under the curve (AUC) of receiver operating characteristics (ROC). RESULTS: The CNN revealed a sensitivity, specificity, and ROC AUC with corresponding 95% confidence intervals (CI) of 95.0% (95% CI 83.5% to 98.6%), 76.7% (95% CI 64.6% to 85.6%), and 0.918 (95% CI 0.866-0.970), respectively. In level I, the dermatologists' management decisions showed a mean sensitivity and specificity of 89.0% (95% CI 87.4% to 90.6%) and 80.7% (95% CI 78.8% to 82.6%). With level II information, the sensitivity significantly improved to 94.1% (95% CI 93.1% to 95.1%; P < 0.001), while the specificity remained unchanged at 80.4% (95% CI 78.4% to 82.4%; P = 0.97). When fixing the CNN's specificity at the mean specificity of the dermatologists' management decision in level II (80.4%), the CNN's sensitivity was almost equal to that of human raters, at 95% (95% CI 83.5% to 98.6%) versus 94.1% (95% CI 93.1% to 95.1%); P = 0.1. In contrast, dermatologists were outperformed by the CNN in their level I management decisions and level I and II diagnostic decisions. More experienced dermatologists frequently surpassed the CNN's performance. CONCLUSIONS: Under less artificial conditions and in a broader spectrum of diagnoses, the CNN and most dermatologists performed on the same level. Dermatologists are trained to integrate information from a range of sources rendering comparative studies that are solely based on one single case image inadequate.

Prognostic implications of baseline anaemia and changes in haemoglobin concentrations with amphotericin B therapy for cryptococcal meningitis.

OBJECTIVES: Anaemia represents a common toxicity with amphotericin B-based induction therapy in HIV-infected persons with cryptococcal meningitis. We sought to examine the impact of amphotericin-related anaemia on survival. METHODS: We used data from Ugandan and South African trial participants to characterize the variation of haemoglobin concentrations from diagnosis to 12 weeks post-diagnosis. Anaemia severity was classified based on the haemoglobin concentration at cryptococcal meningitis diagnosis, and nadir haemoglobin values during amphotericin induction. Cox proportional hazard models were used to estimate 2- and 10-week mortality risk. We also estimated 10-week mortality risk among participants with nadir haemoglobin < 8.5 g/dL during amphotericin induction and who survived ≥ 2 weeks post-enrolment. RESULTS: The median haemoglobin concentration at meningitis diagnosis was 11.5 g/dL [interquartile range (IQR) 9.7-13 g/dL; n = 311] with a mean decline of 4.2 g/dL [95% confidence interval (CI) -4.6 to -3.8; P < 0.001; n = 148] from diagnosis to nadir value among participants with baseline haemoglobin ≥ 8.5 g/dL. The median haemoglobin concentration was 8.1 g/dL (IQR 6.5-9.5 g/dL) at 2 weeks, increasing to 9.4 g/dL (IQR 8.2-10.9 g/dL) by 4 weeks and continuing to increase to 12 weeks. Among participants with haemoglobin < 8.5 g/dL at diagnosis, mortality risk was elevated at 2 weeks [hazard ratio (HR) 2.7; 95% CI 1.5-4.9; P < 0.01] and 10 weeks (HR 1.8; 95% CI 1.1-2.2; P = 0.03), relative to those with haemoglobin ≥ 8.5 g/dL. New-onset anaemia occurring with amphotericin therapy did not have a statistically significant association with 10-week mortality (HR 2.0; 95% CI 0.5-9.1; P = 0.4). CONCLUSIONS: Amphotericin induced significant haemoglobin declines, which were mostly transient and did not impact 10-week mortality. Individuals with moderate to life-threatening anaemia at baseline had a higher mortality risk at 2 and 10 weeks post-enrolment.

Primary non-Hodgkin's lymphoma of vagina associated with pregnancy.

We report a case of non-Hodgkin's lymphoma (NHL) in 23 years old, 36 weeks pregnant female, who presented with a swelling in the vagina. Biopsy of the lesion led to profuse haemorrhage and emergency caesarean-section along with bilateral ligation of internal iliac arteries were required. Recovery was smooth and she was discharged from the hospital on the 5th postoperative day. Histopathology revealed non-Hodgkin's lymphoma, diffuse large B cell type (DLBCL). Further treatment by medical oncologist led to complete resolution of the tumour at three months.

[Entero-mesenteric venous infarction and pregnancy. A recent case]. / Infarctus entéro-mésentérique veineux et grossesse. Une observation récente.

The authors report a case of mesenteric venous infarction in a pregnant woman. After having studied the clinical characteristics of this accident, they discuss the role of pregnancy in its accuracy and also the therapy that has to be adopted.

[Diastematomyelia: antenatal diagnosis with successful outcome, two cases]. / Diastématomyélie: à propos de deux observations anténatales et revue de la littérature.

To discuss, through a review of the literature, clinical signification and antenatal and postnatal management of diastematomyelia. We present two cases highlighting the clinical presentation and discuss management options. The first case was diagnosed in a 32-year-old primagravida female physician. The conssanguinous couple had a history of primary hypofertility. Antenatal magnetic resonance imaging revealed the presence of a sagittal spinal spur separating two asymmetrical hemi cords. Pregnancy was not interrupted due to the lack of factors of poor prognosis. The infant was delivered at term and outcome has been favorable at 18 months follow-up. Ultrasonography performed at 22 weeks gestation provided the diagnosis in the second case. The mother was a 30-year-old primagravida anesthesist who had been addressed for suspected spina bifida. The consanguinous couple had a history of male hypoferility. The ultrasound scan revealed an echogenic spinal spur and integrity of the skin. Pregnancy was not interrupted due to the lack of factors of poor prognosis. The infant was delivered at term and outcome has been favorable at 18 months follow-up. Diastematomyelia is a rare malformation which can be diagnosed antenatally. Careful ultrasonography can distinguish diastematomyelia from myelomeningocele. Antenatal assessment is essential to identify forms with good prognosis because the neonatal outcome of isolated diastematomyelia is generally good, even if surgical repair is required.

Salmonella mediastinitis--a rare disease.

A case report is presented of a young man who was seen because of pleuritic chest pain and fever. CT cross sectional imaging defined a mediastinal mass. Diagnosis of acute mediastinitis due to Salmonella java was made from culturing material at mediastonotomy.

A dose-finding study of raltitrexed (tomudex) with cisplatin and epirubicin in advanced gastro-oesophageal adenocarcinoma.

The standard treatment for advanced gastro-oesophageal cancer in the UK is epirubicin, cisplatin and continuous infusion 5-fluoruracil by an indwelling central venous catheter (ECF), which has significant morbidity. Raltitrexed (tomudex), a specific inhibitor of thymidylate synthase with a long plasma terminal half-life (50-100 h) has activity in gastro-intestinal tract malignancy. To reduce the Hickman line-associated morbidity of ECF; we have conducted a dose-finding study of tomudex combined with epirubicin and cisplatin. Twenty-four patients (22 males, two female), median age 63 years (range 21-75), ECOG performance status < or =2 with histologically proven, unresectable or metastatic gastric (14 patients), gastro-oesophageal junction (nine patients) or oesophageal (one patient) adenocarcinoma received treatment with 3-weekly cisplatin 60 mg m(-2), epirubicin 50 mg m(-2) and tomudex at doses of 2 mg m(-2), 2.5 mg m(-2) or 3 mg m(-2) in successive cohorts. Six patients were treated per dose level with no intra-patient dose escalation. Dose escalation occurred after six patients had completed at least one cycle of chemotherapy at the previous dose level. After defining the maximum tolerated dose a further six patients were treated at the preceding dose level to assess toxicity at the proposed phase II dose. A total of 102 cycles (50% completed 6 cycles) were administered. The dose-limiting toxicities are neutropenia and diarrhoea occurring in 2/6 patients at the 3 mg m(-2) dose level. Of those patients evaluable for response, there were eight partial and one complete response (overall response rate 38%). The median survival was 9.9 months. ECT is an active regimen in oesophagogastric adenocarcinoma. The recommended dose of tomudex for further study in combination with epirubicin and cisplatin is 2.5 mg m(-2).