Simulating mismatch between calibration and target population in AI for mammography the retrospective VAIB study.

Journal: NPJ digital medicine
Published Date:

Abstract

AI cancer detection models require calibration to attain the desired balance between cancer detection rate (CDR) and false positive rate. In this study, we simulate the impact of six types of mismatches between the calibration population and the clinical target population, by creating purposefully non-representative datasets to calibrate AI for clinical settings. Mismatching the acquisition year between healthy and cancer-diagnosed screening participants led to a distortion in CDR between -3% to +19%. Mismatching age led to a distortion in CDR between -0.2% to +27%. Mismatching breast density distribution led to a distortion in CDR between +1% to 16%. Mismatching mammography vendors lead to a distortion in CDR between -32% to + 33%. Mismatches between calibration population and target clinical population lead to clinically important deviations. It is vital for safe clinical AI integration to ensure that important aspects of the calibration population are representative of the target population.

Authors

  • Haiko Schurz
    Department of Oncology-Pathology, Karolinska Institutet, Solna, Sweden. haiko.schurz@ki.se.
  • Klara Solander
    Department of Oncology-Pathology, Karolinska Institutet, Solna, Sweden.
  • Davida Åström
    Department of Oncology-Pathology, Karolinska Institutet, Solna, Sweden.
  • Fernando Cossío
    Department of Oncology-Pathology, Karolinska Institutet, Solna, Sweden.
  • Taeyang Choi
    Department of Oncology-Pathology, Karolinska Institutet, Solna, Sweden.
  • Magnus Dustler
    Department of Translational Medicine, Diagnostic Radiology, Lund University, Malmö, Sweden.
  • Claes Lundström
    Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden.
  • Håkan Gustafsson
    Center for Medical Image Science and Visualization (CMIV), Linköping University, Linköping, Sweden.
  • Sophia Zackrisson
    Department of Translational Medicine, Diagnostic Radiology, Lund University, Malmö, Sweden.
  • Fredrik Strand
    Department of Oncology and Pathology, Karolinska Institute, Stockholm, Sweden.

Keywords

No keywords available for this article.