Automatic determination of cardiovascular risk by CT attenuation correction maps in Rb-82 PET/CT

Abstract

Background: We investigated fully automatic coronary artery calcium (CAC) scoring and cardiovascular disease (CVD) risk categorization from CT attenuation correction (CTAC) acquired at rest and stress during cardiac PET/CT and compared it with manual annotations in CTAC and with dedicated calcium scoring CT (CSCT).

Methods and results: We included 133 consecutive patients undergoing myocardial perfusion ⁸²Rb PET/CT with the acquisition of low-dose CTAC at rest and stress. Additionally, a dedicated CSCT was performed for all patients. Manual CAC annotations in CTAC and CSCT provided the reference standard. In CTAC, CAC was scored automatically using a previously developed machine learning algorithm. Patients were assigned to a CVD risk category based on their Agatston score (0, 1-10, 11-100, 101-400, >400). Agreement in CVD risk categorization between manual and automatic scoring in CTAC at rest and stress resulted in Cohen's linearly weighted κ of 0.85 and 0.89, respectively. The agreement between CSCT and CTAC at rest resulted in κ of 0.82 and 0.74, using manual and automatic scoring, respectively. For CTAC at stress, these were 0.79 and 0.70, respectively.

Conclusion: Automatic CAC scoring from CTAC PET/CT may allow routine CVD risk assessment from the CTAC component of PET/CT without any additional radiation dose or scan time.

Keywords: Automatic calcium scoring; CT attenuation correction map; cardiac CT; cardiovascular risk; coronary calcium.

Figure 1

Manually determined (x-axis) vs automatically computed (y-axis) CAC Agatston scores in A CTAC images at rest and B CTAC images at stress

Figure 2

Examples of calcifications correctly detected by the automatic algorithm in CTAC scans. A CAC lesions in the RCA and LCX in CTAC at rest in a scan with metal implants. B CAC in the RCA strongly affected by cardiac motion in CTAC scan at stress showing severe abnormalities in the lungs

Figure 3

A CAC in LAD missed by the automatic algorithm that resulted in underestimation of CVD risk. The CAC lesion appears blurred, probably due to cardiac motion and large pixel size. B Calcification in the ascending aorta near the right coronary ostium detected as CAC by the automatic method. This large false positive lesion caused overestimation of CVD risk categorization

Figure 4

Manually determined CAC Agatston scores in CSCT (x-axis) vs automatically computed CAC scores in CTAC (y-axis) A at rest and B at stress

Figure 5

One slice from A CSCT, B CTAC at rest and C CTAC at stress of the same patient showing the LAD. The CSCT scan clearly visualizes a coronary artery stent, while the same stent appears as somewhat elongated CAC in the LAD causing large overestimation of the CAC score in CTAC images

Figure 6

One slice from A CSCT, B CTAC at rest and C CTAC at stress of the same patient showing a large CAC in the LCX. While CAC in the CSCT appears large, only some of its voxels exceed 130 HU threshold value leading to substantial CVD risk underestimation (Agatston scores 163, 11, and 6, respectively)