Daily QC is performed using a 40 mm plexiglass block covered the detector area (Figure 1).

Fig 1: Daily quality control (QC) of a mammography equipment with a 40 cm plexiglass block covering the entire detector.

The unprocessed image is automatically analyzed by DOSESTAT QC software, which examines exposure parameters, mean pixel value (MV), signal-to-noise ratio (SNR), and standard deviation (SD). The results and feedback appear in the user interface almost immediately. All data preserved on the medical physicist interface.

For the analysis, the software divides the image into five sections: Top-Left, Top-Right, Bottom-Left, Bottom-Right, and the Breast area which covers the most essential part of automatic exposure control (AEC). The Breast area section presents separately (Figure 2).

Fig 2: Illustration of the ROI map for image analysis, which is divided into five sections, Top-Left, Top-Right, Bottom-Left, Bottom-Right, and the Breast area (AEC area).

Each region of interest (ROI) of approximately 1 cm² has a fixed position. The software identifies each pixel value and calculates the ROI's mean MV, SD, SNR, and percentage deviations from its own section for MV  (Diff MV) and SNR (Diff SNR). See Figure 3.

Fig 3: Illustration of the image analysis. Each ROI of approximately one square centimeter has its fixed position. Calculated mean pixel value (MV), standard deviation (SD), signal-to-noise ratio (SNR) in the ROI visualized. Also the deviation of the ROI's MV and SNR from its own section is shown as Diff MV and Diff SNR in percentage.

Key Improvements

Image Analysis

The image analysis was enhanced by implementing fixed ROI positions and comparing individual pixel values to the mean ROI value. This approach allows for more precise detection of pixel errors and deviations.

SNR was previously found to vary greatly between the breast area and the area outside the nipple due to heel effect and lag or ghosting from previous screening images. Dividing the image analysis into sections improved trend analysis and baseline setting.

Visualization

The SNR visualization on user interface was redesigned using a grayscale characterization that turns out to be specific to each mammography equipment. High SNR values are displayed in dark gray, while low SNR values appear in light gray (Figure 4). This new visualization replaces the previous homogeneous image display.

Fig 4: On the user interface, the analyzed image displayed as a gray-scale characterization of the SNR . High SNR was shown as dark gray on the breast area at the chest wall side of the detector due to heel effect and/or lag/ghosting. The SNR in the breast area (AEC area) was shown separately for an extra attention at possibly pixel error. The staff receives feedback in a second if QC passed or failed. In this case, QC is approved.

The user interface now features an interactive element, allowing users to click on the image and view separate sections of small ROIs with calculated values. If one or more ROI turn pink, which indicated strongly deviating pixels or dead pixels. Red dots in the ROI show the exact location of possible dead pixels.

Exposure Parameter Tuning

Improvements on the medical physicist interface have been made to exposure parameter tuning traceability by displaying the history of changed tolerances (Figure 5 ). This feature enhances the ability to track and analyze equipment performance over time (Figure 6).

Fig 5: Medical Physicist allows setting tolerances for exposure parameters and can freely adjust at any time. The history of what was changed by whom and when is available with a click on history in the medical physicist interface.

Fig 6: Trend analysis on the medical physicist interface. One or more parameters can be visualized with different choices of time intervals.