Narrow Results

Pneumothorax is a common yet potentially serious lung disease, which makes prompt diagnosis and treatment critical in clinical practice. Deep learning methods have proven effective in detecting pneumothorax lesions in medical images and providing quantitative analysis. However, due to the irregular shapes and uncertain positions of pneumothorax lesions, current segmentation methods must be further improved to increase accuracy. This study aimed to propose a Dense Swin-Unet algorithm that integrated the Dense Swin Transformer Block with the Swin-Unet model. The Dense Swin-Unet algorithm employed a sliding window self-attentiveness mechanism on different scales to enhance multiscale long-range dependencies. We designed an enhanced loss function that accelerated the convergence speed to address the issue of class imbalance. Given the limited availability of data in pneumothorax image processing, we created a new dataset and evaluated the efficacy of our model on this dataset. The results demonstrated that our lesion segmentation algorithm attained a Dice coefficient of 88.8%, representing a 1.5% improvement compared with previous deep learning algorithms. Notably, our algorithm achieved a significant enhancement in segmenting small microlesions.

In this paper, we propose a novel registration method based on Adaptive Genetic Algorithm (adaptive-GA) to accomplish the registration task of computer tomography (CT) and its corresponding facial surface data. First, chain code method is utilized to represent the facial curve such that facial surface data obtained by a face scanner is efficiently reduced. Next, based on the concept of the genetic algorithm in continuous space (GACS), we improve its evolutional mechanisms of chromosomes' crossover and mutation to speed up the convergent process of general GA. From the experimental results, it is proved that the proposed registration method can be used for non-fiducials stereotactic brain surgeries and help surgeons to diagnose and treat brain disease correctly and conveniently.

Recently, maximization mutual information (MMI) of image intensities has been proposed as a new matching criterion for automated multimodality image registration. However, the success of the MMI relies on the similarity of the histogram distribution between the images to be fused. This condition is usually hard to be achieved in practical application. Besides, MMI is time consuming because it needs to find an optimal solution about six parameters (three for shifts and three for rotations) during the registration process. To improve the performance of traditional MMI, a novel scheme named Surface-Projection MMI algorithm (SP-MMI) is proposed here. SP-MMI is a two-stage registration algorithm included MMI of surface projection and the following MMI of axial plane. The experimental results, using MR and SPECT images, to confirm the good performance of the proposed method in comparison with the traditional MMI method are also included.