Problem-solving in biological and medical fields using machine learning technique
博士 === 國立交通大學 === 資訊科學與工程研究所 === 106 === The research theme is the use of machine learning technologies with the application in the areas of biological and medical signals/images analysis. We designed a machine learning soma detection algorithm, that effectively identified soma in the raw confocal m...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2018
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/8da5rb |
| Summary: | 博士 === 國立交通大學 === 資訊科學與工程研究所 === 106 === The research theme is the use of machine learning technologies with the application in the areas of biological and medical signals/images analysis. We designed a machine learning soma detection algorithm, that effectively identified soma in the raw confocal microscopic images. Soma has an essential meaning in neuron circuit. We use raw data in FlyCircuit database for an experiment. The F1 score of the proposed method was 85.9%, better than other existing methods. For the medical images analysis, machine learning computer-assisted system was designed for Hepatocellular Carcinoma (HCC liver cancer) diagnosis. The data used were computed tomography and positron emission tomography images. The accuracy was 75% better than existing methods that was 60% to 70%. Finally, we designed wearable devices and applied machine learning method to analyze the signals recorded from the device. Experimental results show that activities and trend/anomaly of gaits can be identified by the proposed methods. All the experimental results show that the machine learning technologies can improve the accuracy.
|
|---|