Related Experiment Video
Updated: Mar 20, 2026
07:53
Author Spotlight: Advancing 3D Modeling for Enhanced Diagnosis and Treatment of Pulmonary Nodules in Early-Stage Lung Cancer
Published on: October 13, 2023
2.2K
Maximum-Intensity-Projection and Computer-Aided-Detection Algorithms as Stand-Alone Reader Devices in Lung Cancer
Lukas Ebner1, Justus E Roos1, Jared D Christensen1
11 Department of Radiology, Duke University Medical Center, Box 3808, Durham, NC 27710.
AJR. American Journal of Roentgenology
|June 2, 2016
Summary
Microdose computed tomography (CT) effectively detects lung nodules with computer-aided detection (CAD) systems, achieving high sensitivity comparable to standard doses. Maximum-intensity-projection (MIP) images also aid detection with fewer false positives.
Area of Science:
- Radiology
- Medical Imaging
- Oncology
Background:
- Lung nodule detection is crucial for early cancer diagnosis.
- Standard-dose computed tomography (CT) carries radiation risks.
- Microdose CT offers a lower-radiation alternative.
Purpose of the Study:
- To evaluate lung nodule detection rates using two computer-aided detection (CAD) systems and maximum-intensity-projection (MIP) images on standard and microdose chest CT.
- To assess the impact of different reconstruction kernels on detection performance.
Main Methods:
- An anthropomorphic chest phantom with 266 nodules (ground-glass and solid) was scanned using standard and microdose CT with three reconstruction kernels (i30, i50, i70).
- Two CAD systems (SyngoCT-CAD [CAD1] and Lung CAD [CAD2]) and MIP images were evaluated by blinded readers.
- Nodule detection sensitivities were calculated for each system, dose, and kernel.
Main Results:
- Microdose CT dose was significantly lower than standard-dose CT (0.1323 mSv vs 1.65 mSv).
- CAD1 demonstrated superior performance over CAD2 across all conditions (p < 0.0001).
- At microdose levels, CAD1 achieved 96.0% sensitivity with the i30 kernel, comparable to stand-alone performance (97.6%).
- Pooled sensitivities for CAD1 ranged from 88.9% to 97.3%, while CAD2 ranged from 49.6% to 73.9%.
- The highest sensitivity (97.3%) was achieved with standard-dose CT, i50 kernel, and CAD1.
Conclusions:
- Sub-millisievert CT is feasible for detecting solid and ground-glass lung nodules.
- Soft-tissue kernels are necessary for optimal performance of MIP and CAD systems.
- MIP reconstructions are valuable adjuncts for increasing sensitivity and reducing false-positive rates in chest CT interpretation.
