Search

Objectives: The diversifying modalities of treatment for gastric cancer raise urgent demands for the rapid and precise diagnosis of metastases in regional lymph nodes, thereby significantly impact the workload of pathologists. Meanwhile, the recent advent of whole-slide scanners and deep-learning techniques have enabled the computer-assisted analysis of histopathological images, which could help to alleviate this impact. Thus, we developed a deep learning-based diagnostic algorithm to detect lymph node metastases of gastric adenocarcinoma and evaluated its performance. Methods: We randomly selected 20 patients with gastric adenocarcinoma who underwent surgery as definitive treatment and were found to be node metastasis-positive. HEMATOXYLIN-eosin (HE) stained glass slides, including a total of 51 metastasis-positive nodes, were retrieved from the specimens of these cases. Other slides with 776 metastasis-negative nodes were also retrieved from other twenty cases with the same disease that were diagnosed as metastasis-negative by the final pathological examinations. All glass slides were digitized using a whole-slide scanner. A deep-learning algorithm to detect metastases was developed using the data in which metastasis-positive parts of the images were annotated by a well-trained pathologist, and its performance in detecting metastases was evaluated. Results: Cross-validation analysis indicated an area of 0.9994 under the receiver operating characteristic curve. Free-response receiver operating characteristic curve (FROC) analysis indicated a sensitivity of 1.00 with three false positives. Further evaluation using an independent dataset also showed similar level of accuracies. Conclusion: This deep learning-based diagnosis-aid system is a promising tool that can assist pathologists involved in gastric cancer care and reduce their workload.

In bronchoschopy, computer vision systems for navigation assistance are an attractive low-cost solution to guide the endoscopist to target peripheral lesions for biopsy and histological analysis. We propose a decoupled deep learning architecture that projects input frames onto the domain of CT renderings, thus allowing offline training from patient-specific CT data. A fully convolutional network architecture is implemented on GPU and tested on a phantom dataset involving 32 video sequences and $$\sim $$ 60k frames with aligned ground truth and renderings, which is made available as the first public dataset for bronchoscopy navigation. An average estimated depth accuracy of 1.5 mm was obtained, outperforming conventional direct depth estimation from input frames by 60%, and with a computational time of $$\le $$ 30 ms on modern GPUs. Qualitatively, the estimated depth and renderings closely resemble the ground truth. The proposed method shows a novel architecture to perform real-time monocular depth estimation without losing patient specificity in bronchoscopy. Future work will include integration within SLAM systems and collection of in vivo datasets.

HD video sequences are widely used in today's multimedia systems and many of these are encoded with H.264 codec. However, it is still challenging to develop a high-performance H.264 encoder because the H.264 encoding process needs a large amount of computations and memory accesses. In this paper, a novel H.264 encoder is described. This encoder can encode video sequences of full HD 60i at double speed. Both MBAFF and Field-Pic structure are supported as coding tool for interlaced video sequences. The memory bandwidths are reduced by using a hierarchical motion estimation method and a pipeline configuration with consideration of MBAFF. The encoder is implemented with 1637K logic gates and 336.5KB on-chip SRAM in the 65nm CMOS technology.

We propose a rate-controllable near-lossless embedded compression algorithm "TLS-1". The algorithm guarantees a selected compression ratio with the smart combination of variable length coding and fixed length coding. It achieves near-lossless image quality under CR = 2. We apply the algorithm to an IP in an HDTV decoder LSI in order to reduce the required external memory capacity and its bandwidth. The LSI is fabricated in a 65nm CMOS technology.

In this paper, we propose a new multi-frame bi-predictive motion compensation method using a temporal linear extrapolation technique. Significant coding gain is obtained by the proposed method, especially for fading or dissolving scenes, which are among the most difficult scenes to compress by conventional coding standards. Moreover, stable improved coding efficiency is obtained, using block by block adaptation combined with multiframe averaging prediction. The number of reference frames for the proposed temporal prediction method is limited to two successive frames which are already coded. And therefore only addition, subtraction and shift operation are needed to generate the prediction signal of both extrapolation and averaging. As for encoding, automatic adaptation is performed without scene characteristics detection, such as fade-in, fade-out, and dissolve. Therefore, additional computational power for both encoder and decoder to use the proposed method is very small.