Morphological and textural descriptors analysis of digital mammograms with radiological findings to support breast cancer detection using artificial neural networks
Description
Data sources for "Morphological and textural descriptors analysis of digital mammograms with radiological findings to support breast cancer detection using artificial neural networks"
Files
Steps to reproduce
* COMPACIDAD_list_APP: The App requests <i>the name of an Excel list<\i> where the first column should indicate the names of the <i> mammograms DICOM for processing<\i> to be analyzed. The mammograms must be stored in the MATLAB path folder. You can analyze the entire list of mammograms or a subset. The results are displayed in the MATLAB workspace and can be exported to an Excel file specified by the user (the name must have at least 5 characters and end with ".xlsx"). <b>Definition<\b>: The Circularity of a mammogram is analyzed, defined as the average 4PiA\(P*P) of several isolevel slices, where A is the maximum area, and P is its perimeter. The isolevel slice is determined by adaptive thresholding of the mammogram. The threshold is defined as the average between the gray level at which the maximum of the histogram inside the breast is obtained and its shoulder (gray level at which a second maximum appears as a peak in the histogram). Several isolevel slices are generated around the threshold to calculate the Circularity for each slice. The Circularity (<i>k<\i>) of the mammogram is defined as the average of the Circularity values of all the slices made. The results for <i>k<\i> and their central tendency values (standard deviation, mode, median, among others) are obtained for each mammogram. * Ro_list_dicom_APP: The Matlab App requests the <i>name of an Excel list</i> with the names of the <i>DICOM for processing</i> mammograms to be analyzed in the first column. The mammograms must be saved in the MATLAB path folder. You can request analysis of the entire mammogram list or a subset. The results are displayed in the MATLAB work window and can be exported to an Excel file specified by the user (name with at least 5 characters and ending in ".xlsx"). Definition: MHWRE is the average height/width ratio of the extremes of an image profile. <i>maxgap</i> determines the height of extremes beyond which they are considered peaks in a profile. Ro is the average of the MHWRE values from 1,000 profiles within the breast image of each mammogram. The Ro results and their central tendency values (standard deviation, mode, median, etc.) are obtained for each mammogram. * Entrena_ANN_seleccionable and Entrena_varios_clasificadores_ANN_con_fitcnet_WSTD: The Matlab APP's requests the name of an Excel file where the results of each mammogram are organized by column, and in the last row, it is indicated with "1" if the histopathology result was breast cancer or "0" if it was benign. In this way, there will be at least 4 data columns per mammogram (r10, r100, k, and histopathology result), and the number of columns will be equal to the number of mammograms to be analyzed. * BENIGNO.rar, BENIGNO-INCAN.rar, CANCER.rar, CANCER-INCAN.rar, contain anonymized DICOM mammograms analyzed (please get in touch with the author).
Institutions
- Instituto Politecnico Nacional Escuela Superior de Medicina