Paced-curriculum distillation with prediction and label uncertainty for image segmentation.

PURPOSE: In curriculum learning, the idea is to train on easier samples first and gradually increase the difficulty, while in self-paced learning, a pacing function defines the speed to adapt the training progress. While both methods heavily rely on the ability to score the difficulty of data samples, an optimal scoring function is still under exploration. METHODOLOGY: Distillation is a knowledge transfer approach where a teacher network guides a student network by feeding a sequence of random samples. We argue that guiding student networks with an efficient curriculum strategy can improve model generalization and robustness. For this purpose, we design an uncertainty-based paced curriculum learning in self-distillation for medical image segmentation. We fuse the prediction uncertainty and annotation boundary uncertainty to develop a novel paced-curriculum distillation (P-CD). We utilize the teacher model to obtain prediction uncertainty and spatially varying label smoothing with Gaussian kernel to generate segmentation boundary uncertainty from the annotation. We also investigate the robustness of our method by applying various types and severity of image perturbation and corruption. RESULTS: The proposed technique is validated on two medical datasets of breast ultrasound image segmentation and robot-assisted surgical scene segmentation and achieved significantly better performance in terms of segmentation and robustness. CONCLUSION: P-CD improves the performance and obtains better generalization and robustness over the dataset shift. While curriculum learning requires extensive tuning of hyper-parameters for pacing function, the level of performance improvement suppresses this limitation.

BEARD-A Potential Donor Site in Grade 6 and Grade 7 Alopecia: A Case Series.

BACKGROUND: Autologous hair transplantation has been the convention in cases of androgenic alopecia. Routinely, the occipital area serves as an ideal donor site. The advent of follicular unit extraction (FUE) has made body and beard hair harvest a possibility. Beard hair, in particular, has been far more sought after than other parts of the body. MATERIALS AND METHODS: A case series of 20 patients have been documented wherein cases with Grade 6 and 7 androgenic alopecia have been treated with beard hair as an adjunct donor site. The local anatomy, procedural technicalities and method of harvesting have been emphasized. The pre, intra and postoperative records have been maintained. DISCUSSION: The advent of FUE paved way for minimal downtime, better cosmesis and less scarring facilitating the possibility of using non-scalp hair in hair restoration, thus increasing the overall donor graft availability. Beard hair has its characteristic differences when compared to the scalp hair. Alongside there exists an array of advantages and disadvantages. CONCLUSION: Minimal complications and potential advantages have encouraged the usage of beard graft in the recent past. In hindsight, beard to scalp transplantation is a worthwhile alternative in cases demanding an expanded source of donor hair which demands further literary contribution.