Search academic texts

Information × Registration Number 0224U032795, (0122U000161) , R & D reports Title Optimization of radiological diagnosis in patients with lymphoma and lung cancer popup.stage_title Оптимізація радіологічної діагностики у хворих на лімфоми і рак легені Head Halaichuk Ihor Yo., д.мед.н. Registration Date 05-12-2024 Organization Ternopil National Medical University named after I. Gorbachevsky of the Ministry of Health of Ukraine popup.description1 Improving the efficiency of differential computed tomography diagnosis of small nodular lung tumors in patients with lung cancer and study the dynamics of changes in bone tissue in patients with diffuse large B-cell lymphoma popup.description2  Research report: 45 p., 5 tables, 17 figures, 40 sources. OSTEOPOROSIS; BONE MINERAL DENSITY; COMPUTED TOMOGRAPHY; LYMPHOMA; CHEMOTHERAPY; PROGNOSIS; LUNG CANCER; FRACTAL ANALYSIS; LIVER METASTASE. The object of the study is computed tomograms of patients with lung cancer, lymphoma and liver metastases. Objective: to substantiate the use of fractal analysis and mathematical prediction methods to optimize radiological diagnostics in patients with lymphoma and lung cancer and in patients with metastatic liver disease. The study included 55 patients with stage II-III-IV lung cancer aged 45–65 years. According to the recommendations of the Fleischner Society, the dynamics of the doubling of the selected nodules in size, volume and time were monitored, on the basis of which two cohorts of objects for fractal research were formed. The “benign” cohort (n=120 nodules) was formed from pulmonary nodules with stable CT characteristics. The “malignant” cohort (n=75 nodules) was formed from other nodules that underwent changes in CT dynamics, i.e. metastatic transformation. The methods of nonlinear mathematics showed significant differences in fractal dimension between pulmonary nodules of benign and malignant origin. Binary classifiers demonstrated the qualitative work of the proposed research algorithm: decision trees (Classification tree) and k-nearest neighbors (Nearest Neighbor) distinguish benign changes and malignant pathology by 100%. Product Description popup.authors popup.nrat_date 2024-12-05 Close