Background <p>Multi-pinhole cardiac-dedicated CZT gamma cameras improve clinical workflow by reducing acquisition duration and/or injected activities. As most of these systems lack integrated CT scanner, attenuation artifacts may appear in the reconstructed image depending on patient morphology. External CT-based attenuation correction is possible but requires manual SPECT-CT registration and increases patient absorbed dose. The aim of the present study was to develop an approach to estimate patient-specific attenuation maps directly from emission data acquired on a pinhole cardiac CZT camera.</p> Methods <p>Two anthropomorphic torso phantoms were used. Their heart wall (with or without defects), liver insert and background cavity were filled with ^99mTc solutions to simulate clinical activities. Reference acquisitions were obtained using a CZT SPECT/CT system. For each phantom, two acquisitions were performed on a pinhole cardiac CZT camera: a standard cardiac acquisition and a second acquisition with detectors radially moved away. Combination of both acquisitions enabled delineation of the body contour and segmentation of soft-tissue and lung regions. From this segmentation, a CT-free attenuation map was generated and imported into the manufacturer’s reconstruction software to produce attenuation corrected (AC) images. These CT-free AC images were semi-quantitatively compared, using 17-segment polar maps, to (1) images corrected using a registered external CT, (2) AC images from the CZT SPECT/CT system, and (3) the true activity distribution in the phantoms. A three-dimensional voxel-wise comparison was also performed between CT-free AC and CT-based AC images of the pinhole CZT camera.</p> Results <p>CT-free attenuation maps produced tracer distribution and 17-segment polar maps closely matching those obtained with CT-based attenuation correction. Heart wall defects in the anterior and inferior areas remained clearly visible except for half-obstructive defects. Differences relative to CZT SPECT/CT images were mainly attributable to camera geometry and reconstruction algorithm.</p> Conclusions <p>The novel CT-free patient-specific attenuation correction method yields images comparable to CT-based AC images and shows good agreement with AC images from a CZT SPECT/CT system. This approach could enhance the diagnostic performance of current cardiac pinhole CZT cameras without requiring additional CT acquisitions.</p>