This article is a guide to convolutional neural network technologies and their clinical applications in the analysis of radiologic images.

https://pubs.rsna.org/doi/pdf/10.1148/radiol.2018180547

Convolutional Neural Networks for Radiologic Images: A Radiologist’s Guide

Deep learning with a convolutional neural network (CNN) is gaining attention recently for its high performance in image recognition. Images themselves can be utilized in a learning process with this technique, and feature extraction in advance of the learning process is not required. Important features can be automatically learned. Thanks to the development of hardware and software in addition to techniques regarding deep learning, application of this technique to radiological images for predicting clinically useful information, such as the detection and the evaluation of lesions, etc., are beginning to be investigated. This article illustrates basic technical knowledge regarding deep learning with CNNs along the actual course (collecting data, implementing CNNs, and training and testing phases). Pitfalls regarding this technique and how to manage them are also illustrated. We also described some advanced topics of deep learning, results of recent clinical studies, and the future directions of clinical application of deep learning techniques.

Deep learning with convolutional neural network in radiology

https://repositorio-aberto.up.pt/bitstream/10216/124569/3/369051.1.pdf

Deep-learning framework to detect lung abnormality - A study with chest X-Ray and lung CT scan images

Survey on deep learning for radiotherapy.

Hybrid deep learning for detecting lung diseases from X-ray images

Lung cancer is the most common cancer that cannot be ignored and cause death with late health care. Currently, CT can be used to help doctors detect the lung cancer in the early stages. In many cases, the diagnosis of identifying the lung cancer depends on the experience of doctors, which may ignore some patients and cause some problems. Deep learning has been proved as a popular and powerful method in many medical imaging diagnosis areas. In this paper, three types of deep neural networks (e.g., CNN, DNN, and SAE) are designed for lung cancer calcification. Those networks are applied to the CT image classification task with some modification for the benign and malignant lung nodules. Those networks were evaluated on the LIDC-IDRI database. The experimental results show that the CNN network archived the best performance with an accuracy of 84.15%, sensitivity of 83.96%, and specificity of 84.32%, which has the best result among the three networks.

/pdf/using-deep-learning-for-classification-of-lung-nodules-on-3uvmuasu29.pdf

Using Deep Learning for Classification of Lung Nodules on Computed Tomography Images.

To achieve real-time object detection tasks with high throughput and low latency, this paper proposes a multi-die hardware accelerator architecture. It implements three accelerators on the VU9P chip, each of which is bound to an independent super logic region (SLR). To reduce off-chip memory access and power consumption, this design uses three on-chip buffers to store the weights and intermediate result data on one hand; on the other hand, it minimizes data access and movement and maximizes data reuse. This design uses an 8-bit quantization strategy for both weights and feature maps to achieve twice the throughput and computational efficiency of a single digital signal processor (DSP). In addition, many operators are designed in the accelerator, and all of them are fully parameterized, so it is easy to extend the network, and the control of the accelerator can be realized by configuring the instruction group. By accelerating the YOLOv4-tiny algorithm, the accelerator architecture can achieve a frame rate of 148.14 frames per second (FPS) and a peak throughput of 2.76 tera operations per second (TOPS) at 200 MHz with an energy efficiency ratio of 93.15 GOPS/W. The code can be found at https://github.com/19801201/Verilog_CNN_Accelerator.

https://ieeexplore.ieee.org/ielx7/6287639/6514899/09858118.pdf

Design and Implementation of Convolutional Neural Networks Accelerator Based on Multidie

In the general case, the performance of deep learning-based classification models depends on the ability of capturing features. When a sample has various appearances, the increased features may lower the performance of these models. In this case, training more models on different appearances can be a choice to improve the accuracy. In this paper, we built a new framework that generates a network of models to improve the accuracy. First, our framework built a strategy to increase the number of models to well capture the increased features. We then utilise our recursive Bayesian methods on the selected outputs of trained models, which is to reduce the similarity among these outputs for higher accuracy. The experimental results show that our framework can be a good choice to improve the performance of deep learning applications. 

Improving the performance of deep learning-based classification when a sample has various appearances

Object detection from remote sensing images is an interesting research topic full of theoretical and application values. Compared with generic images, those of remote sensing contain much less distinguishing representation information due to far-away capturing position and limited camera lens resolution, the perpendicular to ground capturing view, almost unacquirable 3-D object structure and serious roof occlusion, and the inherent unimportance of top roof view. Focusing on this intrinsic disadvantage, in this letter, we propose a general framework of accurate remote sensing object localization constructed with the core idea of maximum image feature exploration. This integrated structure is mainly comprised of cascaded tiny patch correlation (CTPC) based feature digging, averaging local patch regression (ALPR) centered coarse location acquisition, and instance segmentation oriented further refinement (ISOFR). All those components are designed to maximize feature digging and usage. We believe this is the first to implement the idea of maximum feature exploration to each corner of object detection architecture. Extensive experiments on challenging aerial image datasets DOTA and NWPU VHR-10 show state-of-the-art performance.

A Framework of Maximum Feature Exploration Oriented Remote Sensing Object Detection

Currently, many applications implement convolutional neural networks on CPUs or GPUs while the performances are limited by the computational complexity of these networks. Compared with the implementation on CPUs or GPUs, deploying convolutional neural accelerators on FPGAs can achieve superior performance. On the other side, the multiplication operations of convolutional neural networks have been a constraint for FPGAs to achieve the better performance. In this paper, we proposed a shift convolutional neural network accelerator, which converts the multiplication operations into shift operations. Based on the shift operation, our accelerator can break the computational bottleneck of FPGAs. On Virtex UltraScale+ VU9P, our accelerator can save DSP resources and reduce memory consumption while achieving a performance of 1.18 TOPS(Tera Operations Per Second), which is an essential improvement over the other convolutional neural accelerators. 

QingZeng Song

Papers

Using Deep Learning for Classification of Lung Nodules on Computed Tomography Images.

Design and Implementation of Convolutional Neural Networks Accelerator Based on Multidie

Improving the performance of deep learning-based classification when a sample has various appearances

A Framework of Maximum Feature Exploration Oriented Remote Sensing Object Detection

Design and Implementation of a Universal Shift Convolutional Neural Network Accelerator