Original Article
Radiomics for the detection of diffusely impaired myocardial perfusion: A proof-of-concept study using 13N-ammonia positron emission tomography

https://doi.org/10.1007/s12350-022-03179-yGet rights and content

Abstract

Aim

The current proof-of-concept study investigates the value of radiomic features from normal 13N-ammonia positron emission tomography (PET) myocardial retention images to identify patients with reduced global myocardial flow reserve (MFR).

Methods

Data from 100 patients with normal retention 13N-ammonia PET scans were divided into two groups, according to global MFR (i.e., < 2 and ≥ 2), as derived from quantitative PET analysis. We extracted radiomic features from retention images at each of five different gray-level (GL) discretization (8, 16, 32, 64, and 128 bins). Outcome independent and dependent feature selection and subsequent univariate and multivariate analyses was performed to identify image features predicting reduced global MFR.

Results

A total of 475 radiomic features were extracted per patient. Outcome independent and dependent feature selection resulted in a remainder of 35 features. Discretization at 16 bins (GL16) yielded the highest number of significant predictors of reduced MFR and was chosen for the final analysis. GLRLM_GLNU was the most robust parameter and at a cut-off of 948 yielded an accuracy, sensitivity, specificity, negative and positive predictive value of 67%, 74%, 58%, 64%, and 69%, respectively, to detect diffusely impaired myocardial perfusion.

Conclusion

A single radiomic feature (GLRLM_GLNU) extracted from visually normal 13N-ammonia PET retention images independently predicts reduced global MFR with moderate accuracy. This concept could potentially be applied to other myocardial perfusion imaging modalities based purely on relative distribution patterns to allow for better detection of diffuse disease.

Introduction

Nuclear myocardial perfusion imaging (MPI) is well established for the assessment of suspected or known coronary artery disease (CAD) and is widely implemented in clinical practice. Single-photon emission computed tomography (SPECT), however, is inherently limited in detecting diffuse myocardial perfusion abnormalities, as this technique relies on normally perfused myocardium as a reference for normalization of myocardial radionuclide retention. This becomes relevant, for example, in a setting of multi-vessel CAD and/or microvascular disease, explaining the moderate sensitivity of SPECT in these clinical situations.1,2 Contrary to conventional SPECT, positron emission tomography (PET) imaging allows for absolute quantification of myocardial perfusion, thus overcoming the above-mentioned shortcomings of SPECT.3,4 Myocardial blood flow (MBF) and myocardial flow reserve (MFR), assessed with PET, provide a quantitative measure integrating the hemodynamic consequences of focal lesions, diffuse lesions, small vessel disease, and microvascular dysfunction on tissue perfusion.

Despite the obvious advantages of PET MPI, its wide implementation is currently hampered by limited availability due to relatively high up-front costs and the need for an on-site cyclotron or generator due to the short half-life of currently available PET perfusion tracers. On the other hand, while newest-generation SPECT cameras with cadmium-zinc-telluride-base detector technology have been shown to allow for dynamic image acquisition, enabling accurate MBF quantitation,5, 6, 7 their availability currently remains limited as well. Thus, the majority of cardiac nuclear studies for evaluating CAD is currently performed on conventional SPECT cameras, which do not allow for quantification of MBF and are prone to false negative findings in patients with balanced multi-vessel disease, left main coronary artery disease or microvascular disease, which are associated with an increased cardiovascular risk.8,9 It is against this background that substantial efforts have been made to identify imaging features within the SPECT data itself to detect the presence of balanced ischemia or coronary microvascular disease. Various approaches have, however, demonstrated moderate performance at best, not qualifying them as robust and reliable markers in clinical routine for the detection of diffusely impaired myocardial perfusion.10, 11, 12

By contrast, the concept of radiomics as a novel supportive approach for image analysis in cardiac nuclear imaging is promising in this regard, as it allows assessing a multitude of subtle image features, which are imperceptible to the naked human eye and may potentially be more beneficial than previously addressed markers. Radiomics and machine learning applications in cardiology and cardiovascular imaging are now quite common in research13 and a number of studies have already demonstrated a potential clinical value of radiomics in nuclear medicine and radiology, including cardiac CT.14,15 However, the literature on its potential value specifically in SPECT MPI remains scarce.

We hypothesize that the application of radiomics to myocardial perfusion radionuclide retention images yields diagnostic value for the detection of diffusely impaired perfusion. The current work aims to provide proof-of-concept using 13N-ammonia PET MPI. Contrary to SPECT MPI, the latter provides the unique advantage of providing both the retention images and the standard of truth through absolute MBF quantification.

Section snippets

Patient collection

This is a retrospective single-center matched cohort study including patients with normal retention images on 13N-ammonia PET MPI scans. We identified patients from the Zurich Quantitative PET Registry16 with preserved and decreased global MFR as derived from PET quantification, matched by age, gender, and body mass index. Preserved MFR was defined as global MFR ≥ 2 and decreased MFR was defined as global MFR < 2.17 This study was approved by the local ethics committee (BASEC-Nr. 2016-09177).

PET acquisition, reconstruction, and analysis

Patient characteristics

Fifty patients with decreased MFR (group 1) and 50 patients matched by gender, age and body mass index with preserved MFR (group 2) were randomly selected from the Zurich Quantitative PET Registry, generating a cohort of 100 patients. The patient population consisted of 58 females (58%) and 42 males (42%) with an average age of 64 ± 12 years. Average global MFR in group 1 and group 2 was 1.74 ± 0.18 and 3.20 ± 0.81, respectively (P < .0001). A validation cohort of 30 additional patients was

Discussion

In this proof-of-concept study, using 13N-ammonia PET MPI, we demonstrate the value of radiomic features extracted from normal retention images to identify patients with globally reduced MFR.

The principle of radiomics analysis builds on the extraction of voxel values, their relationship to each other as well as texture patterns from the image, providing quantitative characteristics of medical images which are imperceptible to the human eye. Radiomics have been extensively studied in

Conclusion

A single radiomic feature (GLRLM_GLNU) extracted from visually normal 13N-ammonia PET retention images independently predicts reduced global MFR with moderate accuracy. This concept could potentially be applied to other myocardial perfusion imaging modalities based purely on relative distribution patterns to allow for better detection of diffuse disease.

New knowledge gained

Radiomics analysis from 13N-ammonia PET retention images is feasible and a single radiomic predicts globally reduced MFR in patients with visually normal 13N-ammonia PET retention images.

Funding

Open access funding provided by University of Zurich.

Disclosures

All authors report no relationships that could be construed as a conflict of interest.

References (29)

  • ZavadovskyKV et al.

    The current status of CZT SPECT myocardial blood flow and reserve assessment: Tips and tricks

    J Nucl Cardiol

    (2021)
  • FujimotoS et al.

    Study of the predictors and lesion characteristics of ischemic heart disease patients with false negative results in stress myocardial perfusion single-photon emission tomography

    Circ J

    (2006)
  • FangY-HD et al.

    Detecting triple-vessel disease with cadmium zinc telluride-based single-photon emission computed tomography using the intensity signal-to-noise ratio between rest and stress studies

    Contrast Media Mol Imaging

    (2017)
  • DriessenRS et al.

    Adenosine single-photon emission computed tomography-derived transient ischemic dilatation and ejection fraction reserve fail to predict multivessel coronary artery disease

    Nucl Med Commun

    (2019)
  • Cited by (0)

    The authors of this article have provided a PowerPoint file, available for download at SpringerLink, which summarises the contents of the paper and is free for re-use at meetings and presentations. Search for the article DOI on SpringerLink.com.

    The authors have also provided an audio summary of the article, which is available to download as ESM, or to listen to via the JNC/ASNC Podcast.

    Ganna Degtiarova and Chrysoula Garefa have contributed equally to this work.

    View full text