This study developed a model to automatically detect blurred areas in mammograms, which can affect diagnostic accuracy. Using a retrospective dataset consisting of 152 mammograms from three vendors, expert radiologists outlined blurred regions. Normalized Wiener spectra (nWS) were extracted and processed through a convolutional neural network (CNN) to classify images as either blurred or sharp. The model showed an AUROC of 0.808, with 78% agreement on blurred mammograms and 75% on sharp ones. The results suggest that frequency-based feature extraction can eliminate subjectivity in mammogram assessments, offering a more reliable tool for radiologists to improve diagnostic accuracy.
Key points:
- Blurring in mammography limits radiologist interpretation and diagnostic accuracy.
- This objective blur detection tool ensures image quality, and reduces retakes and unnecessary exposures.
- Wiener spectrum analysis and CNN enabled automated blur detection in mammography.
Authors: S. Nowakowska, V. Vescoli, T. Schnitzler, C. Ruppert, K. Borkowski, A. Boss, C. Rossi, B. Wein & A. Ciritsis
