Is training essential for interpreting cardiac computed tomography?

BACKGROUND: Cardiac computed tomography (CT) has gained increasing acceptance for diagnosing obstructive coronary artery disease (CAD). Several guidelines have been published on required education for proficiency in the interpretation of these examinations. PURPOSE: To describe the learning-curve effect of the interpretation of 100 consecutive cardiac CT examinations aimed at diagnosing CAD. The diagnostic accuracy of radiologists and radiographers was also compared. MATERIAL AND METHODS: Two radiologists and two radiographers, all with no prior experience in evaluation of cardiac CT, independently underwent a dedicated training program of 100 examinations randomized into 10 blocks (sessions), with 10 cases in each. They independently evaluated the coronary arteries regarding significant obstructive CAD. After every session, individual feedback on diagnostic accuracy and comparison with the corresponding invasive coronary angiography (currently regarded as the gold standard to detect coronary lesions) was given. The time required for interpretation was recorded. RESULTS: The mean review time decreased (P<0.0001) successively during the 10 sessions for all the observers together. The first session had a mean review time of 32 min, and the last session 16 min. No significant improvement in sensitivity, specificity, or negative predictive value (NPV) was observed. For positive predictive value (PPV), there was an improvement for the radiologists (P<0.05), but not for the radiographers. The radiographers had a higher total specificity compared to the radiologists (P<0.01). CONCLUSION: The review time for novices in cardiac CT was approximately halved during the first 100 cases, with maintained accuracy. There was a learning-curve effect in PPV for the radiologists. The diagnostic accuracy of dedicated radiographers indicates that they might be considered to be included as part of the evaluation team.

Limitations of 64-detector-row computed tomography coronary angiography: calcium and motion but not short experience.

BACKGROUND: Recently, 64-detector-row computed tomography coronary angiography (CTA) has been introduced for the noninvasive diagnosis of coronary artery disease. PURPOSE: To evaluate the diagnostic capacity and limitations of a newly established CTA service. MATERIAL AND METHODS: In 101 outpatients with suspected coronary artery disease, 64-detector-row CTA (VCT Lightspeed 64; GE Healthcare, Milwaukee, Wisc., USA) was performed before invasive coronary angiography (ICA). The presence of >50% diameter coronary stenosis on CTA was rated by two radiologists recently trained in CTA, and separately by an experienced colleague. Diagnostic performance of CTA was calculated on segment, vessel, and patient levels, using ICA as a reference. Segments with a proximal reference diameter <2 mm or with stents were not analyzed. RESULTS: In 51 of 101 patients and 121 of 1280 segments, ICA detected coronary stenosis. In 274 of 1280 (21%) segments, CTA had non-diagnostic image quality, the main reasons being severe calcifications (49%), motion artifacts associated with high or irregular heart rate (45%), and low contrast opacification (14%). Significantly more women (43%) had non-diagnostic scans compared to men (20%). A heart rate above 60 beats per minute was associated with significantly more non-diagnostic patients (38% vs. 18%). In the 1006 diagnostic segments, CTA had a sensitivity of 78%, specificity of 95%, positive predictive value (PPV) of 54%, and negative predictive value (NPV) of 98% for detecting significant coronary stenosis. In 29 patients, CTA was non-diagnostic. In the remaining 72 patients, sensitivity was 100%, specificity 65%, PPV 79%, and NPV 100%. The use of a more experienced CTA reader did not improve diagnostic performance. CONCLUSION: CTA had a very high negative predictive value, but the number of non-diagnostic scans was high, especially in women. The main limitations were motion artifacts and vessel calcifications, while short experience in CTA did not influence the interpretation.