Discrete-space versus continuous-space lesion boundary and area definitions

Area of Science:

• Medical imaging analysis
• Quantitative image analysis
• Biomedical image processing

Background:

• Accurate measurement of anatomic regions in medical images is vital for disease diagnosis, growth tracking, and therapy response evaluation.
• Medical images, being discrete, allow for various boundary definitions and area calculation methods.
• Discrepancies in these methods can lead to substantially different quantitative values.

Purpose of the Study:

• To investigate the impact of different boundary definitions and area calculation methods on the measurement of anatomic regions in medical images.
• To compare the accuracy of various region boundary definitions (continuous polygon, internal discrete, external discrete) and area calculation techniques (pixel counting, Green's theorem).

Main Methods:

• Applied multiple boundary definitions and area calculation methods to synthetic and clinical image databases.
• Utilized the Lung Image Database Consortium (LIDC) database for lung nodules and a database of adrenal gland outlines.
• Compared quantitative area measurements obtained from different methodological approaches.

Main Results:

• Significant differences in area measurements, averaging around 20%, were observed among the various methods applied to clinical databases.
• Inconsistent application of region boundary definitions and area calculation methods substantially impacts measurement accuracy.
• The choice of method significantly affects the quantitative assessment of anatomic regions.

Conclusions:

• The study highlights the critical need for standardized methodologies in defining region boundaries and calculating areas in medical imaging.
• Inconsistent measurement techniques can introduce substantial errors, potentially affecting clinical decisions.
• Standardization is essential for improving the reliability and reproducibility of quantitative medical image analysis.