| Abstrakt: |
This study uses medical infrared imaging to analyze quantitative and qualitative data in an effort to identify breast cancer. One of the biggest benefits of the suggested health procedure is that medical infrared imaging does not include any dangerous radiation. The finest diagnostic parameters among those that are accessible are chosen through analysis of this data, and their sensitivity and precision in the detection of cancer are increased by using machine learning algorithms.A growing number of areas are now adopting thermal imaging cameras as a result of the popularity of photography and video cameras. Thermal imaging can be used in the noninvasive, low-cost diagnosis of breast cancer, which is one such use. Additionally, exposure to high radiation doses, which may be employed in conventional ways of finding breast cancer, might raise one's chance of developing cancer. This essay describes a technique for finding breast cancer using Thermal Images with J48 and Voted Perceptron algorithms. Materials and Methods: Thermal pictures from Visual Labs DMR-IR were used to gather the data for this investigation. The study photographs are all JPEG files with a resolution of 100 dpi and a resolution of 640 x 480 pixels. Each group's sample size is 60, for a total sample size of 120. For J48 and Voted Perceptron, the sample sizes were (N=60) and (N=60), respectively. The accuracy, specificity, and sensitivity calculations are performed using novel Matlab software. Results: J48 had an accuracy of 90.50% before adding the clinical data choice, and this accuracy rose to 90.83% after that. Voted Perceptron had an accuracy of 86%before adding the clinical data choice, and this accuracy rose to 86.5% after that. When ill patients are taken into account as the positive class, the model was able to accurately identify more patients with the addition of clinical data judgements. Two tailed value at p=.001(p<0.05) is statistically significant. Conclusion: Utilizing 120 records and 2 clinical characteristics from a dataset, machine learning models were used to examine the survival of breast cancer patients. J48 demonstrated the maximum accuracy (90%) sensitivity(91.6%) specificity (90%) in this novel investigation with Voted Perceptron accuracy(86%)sensitivity (76.6%) specificity (96.6%). [ABSTRACT FROM AUTHOR] |
