Accurate assessment of renal histopathology is crucial for the clinical management of patients with lupus nephritis (LN). However, the current classification system has poor interpathologist agreement. This paper proposes a deep convolutional neural network (CNN)-based system that detects and classifies glomerular pathological findings in LN. A dataset of 349 renal biopsy whole-slide images (WSIs) (163 patients with LN, periodic acid-Schiff stain, 3906 glomeruli) annotated by three expert nephropathologists was used. The CNN models YOLOv4 and VGG16 were employed to localise the glomeruli and classify glomerular lesions (slight/severe impairments or sclerotic lesions). An additional 321 unannotated WSIs from 161 patients were used for performance evaluation at the per-patient kidney level. The proposed model achieved an accuracy of 0.951 and Cohen’s kappa of 0.932 (95% CI 0.915–0.949) for the entire test set for classifying the glomerular lesions. For multiclass detection at the glomerular level, the mean average precision of the CNN was 0.807, with ‘slight’ and ‘severe’ glomerular lesions being easily identified (F1: 0.924 and 0.952, respectively). At the per-patient kidney level, the model achieved a high agreement with nephropathologist (linear weighted kappa: 0.855, 95% CI: 0.795–0.916, p < 0.001; quadratic weighted kappa: 0.906, 95% CI: 0.873–0.938, p < 0.001). The results suggest that deep learning is a feasible assistive tool for the objective and automatic assessment of pathological LN lesions.

BACKGROUND:Systemic lupus erythematosus (SLE) leads to renal lesions, which may be clinically silent in patients with little or no proteinuria. Early detection of these lesions may improve prognosis, but early markers are controversial. This study ai...

Ankylosing spondylitis (AS) is most common in adolescents and the ultimate result is disability, which places a huge burden on patients and society. Therefore, the key to improve the prognosis of AS is the early diagnosis of hip injury. To examine if AS patients whose hip pain is either absent or minimal might already have observable MRI and X-ray hip changes. Clinical and imaging hip data were systematically analyzed in 200 healthy controls (HC) and 300 AS with varying degrees of hip pain. Forty-four patients with early hip osteoarthritis (OA) served as positive imaging controls. In MRI images, BME lesions in the STIR sequence were much more frequent in AS (62%) compared to HC (2%) (p < 0.0001). Most importantly, 42% of AS with no or minimal hip pain had one or more MRI lesions. This was much more frequent compared to the 2% in HC (p < 0.05). These lesions in AS were observed singly or in combination in the trochanters (8%), femoral heads (12%), and acetabula (13%). Parallel finding that X-ray changes were present in patients with minimal or no hip pain was also observed with X-ray. Based on the normal hip width of HC, joint space narrowing was observed in 94.3% of the entire AS cohort, and importantly 56.7% of AS patients with no or mild hip pain. In these latter patients, functional activities of the hips such as walking were normal. At least 40% of AS patients with minimal or no hip pain might already show MRI and X-ray changes.

Objective: The literature describing follow-up carotid contrast-enhanced ultrasonography (CEUS) is limited. We report our experience with monitoring CEUS that is performed in patients with Takayasu arteritis (TAK). Methods: We retrospectively analyzed patients with TAK who had undergone carotid CEUS 2 or more times with a follow-up duration of 12 or more months at Xijing Hospital between 2017 and 2020. We described how CEUS interpretation changed, and we recorded the state of remission (ie, bilateral CEUS visual grades ≤ 1) or relapse as determined by imaging. Results: In total, 106 patients with TAK and 425 CEUS visits were included in the study; the median follow- up was 25 (IQR 18-30) months. The CEUS vascularization grade was significantly associated with the Kerr criteria (r = 0.532, P < 0.001), erythrocyte sedimentation rate (P < 0.001), and C-reactive protein level (P < 0.001). At baseline, 76 patients (71.7%) had active disease as determined by CEUS and 30 (28.3%) had inactive disease. The midterm assessment (median 13, IQR 10-16 months) showed that 29 out of 76 CEUS-active patients (38.2%) achieved complete response, 34 (44.7%) achieved partial response, and 13 (17.1%) did not respond. At the last visit, the total number of responders was 78 out of 94 (83.0%). CEUS relapse was observed in 28 out of 57 (49.1%) patients, with a median of 16 (IQR 10-21) months. The Kaplan-Meier curve demonstrated that the remission rate evaluated by the CEUS-combined method (median 22 months) was lower than that of the clinical-only evaluation (median 11 months; P < 0.001). Conclusion: Response or relapse according to CEUS was detected in most patients during follow-up. CEUS is an effective technique for detecting carotid artery inflammation in patients with TAK.