Tissue at Risk and Ischemic Core Estimation Using Deep Learning in Acute Stroke.

BACKGROUND AND PURPOSE: In acute stroke patients with large vessel occlusions, it would be helpful to be able to predict the difference in the size and location of the final infarct based on the outcome of reperfusion therapy. Our aim was to demonstrate the value of deep learning-based tissue at risk and ischemic core estimation. We trained deep learning models using a baseline MR image in 3 multicenter trials. MATERIALS AND METHODS: Patients with acute ischemic stroke from 3 multicenter trials were identified and grouped into minimal (≤20%), partial (20%-80%), and major (≥80%) reperfusion status based on 4- to 24-hour follow-up MR imaging if available or into unknown status if not. Attention-gated convolutional neural networks were trained with admission imaging as input and the final infarct as ground truth. We explored 3 approaches: 1) separate: train 2 independent models with patients with minimal and major reperfusion; 2) pretraining: develop a single model using patients with partial and unknown reperfusion, then fine-tune it to create 2 separate models for minimal and major reperfusion; and 3) thresholding: use the current clinical method relying on apparent diffusion coefficient and time-to-maximum of the residue function maps. Models were evaluated using area under the curve, the Dice score coefficient, and lesion volume difference. RESULTS: Two hundred thirty-seven patients were included (minimal, major, partial, and unknown reperfusion: n = 52, 80, 57, and 48, respectively). The pretraining approach achieved the highest median Dice score coefficient (tissue at risk = 0.60, interquartile range, 0.43-0.70; core = 0.57, interquartile range, 0.30-0.69). This was higher than the separate approach (tissue at risk = 0.55; interquartile range, 0.41-0.69; P = .01; core = 0.49; interquartile range, 0.35-0.66; P = .04) or thresholding (tissue at risk = 0.56; interquartile range, 0.42-0.65; P = .008; core = 0.46; interquartile range, 0.16-0.54; P < .001). CONCLUSIONS: Deep learning models with fine-tuning lead to better performance for predicting tissue at risk and ischemic core, outperforming conventional thresholding methods.

Survivin in adrenocortical tumors - pathophysiological implications and therapeutic potential.

Treatment options for adrenocortical carcinoma (ACC) are very limited. In other solid tumors, small vaccination trials targeting the anti-apoptotic molecule survivin suggested immunological and clinical benefit in selected patients. Therefore, we investigated whether survivin might be a suitable target for immunotherapy in ACC. Survivin mRNA and protein expression was assessed in adrenal tissue specimens [by real-time-PCR in 29 ACC, 24 adrenocortical adenomas (ACA) and 12 normal adrenal glands; by immunohistochemistry in 167 ACCs, 15 ACA, and 5 normal adrenal glands]. Expression was correlated with clinical outcome using Kaplan-Meier and Cox regression analyses. The anti-apoptotic role of survivin was investigated in the SW13 ACC cell line using survivin siRNA. The presence of spontaneous survivin specific T-cells in peripheral blood was assessed by FACS dextramere staining in 29 ACC patients in comparison to healthy controls. Survivin mRNA in ACC was significantly overexpressed when compared with ACA or normal adrenal glands. Immunohistochemistry confirmed survivin protein expression in 97% of the ACCs. In 83% of samples, staining was moderate or high and clinical outcome in this subgroup showed a trend towards poorer prognosis [hazard ratio for death 2.28 (95% CI 0.99-5.28); p=0.053]. Survivin knockdown in SW-13 cell significantly increased the rate of apoptosis. Finally, spontaneous survivin-reactive T cells were detectable in 3 of 29 ACC patients. In conclusion, our data suggest that survivin could play an important role in the anti-apoptotic mechanisms in ACC and provide first hints that targeting survivin might be an interesting new therapeutic approach in this rare disease.

Comparative investigations of structure and properties of BCN coatings deposited by thermal and plasma-enhanced CVD.

Hard BCN films were deposited by chemical vapour deposition (CVD) on Si(100) substrates. TCVD (thermal activated) and PECVD (GD or RF microwave plasma-activated, respectively) were used. The films were analysed with respect to chemical state, composition, morphology and microstructure, oxidation behaviour and hardness. Wavelength dispersive X-ray spectroscopy (EPMA), infrared spectroscopy (IR), transmission electron microscopy (TEM), differential thermal analysis (DTA) and hardness evaluation were employed for film characterization. A correlation between deposition parameters and film composition, structure and hardness could be proved in every CVD process. Parallels between TCVD and PECVD films emerged in the case of chemical composition and the correlation between carbon content and hardness values. Considerable differences exist with regard to the microstructure, especially the texture of the films. Moreover in TCVD films the carbon is preferentially incorporated between the BN basal planes, whereas in PECVD films it is incorporated preferentially in as well as between the BN basal planes.