Radiomics-based left ventricular ejection fraction prediction: a feasibility study

doi:10.3969/j.issn.1674-8115.2023.09.010

Abstract

Abstract:

Objective ·To assess the feasibility of using 3D imaging features extracted from cardiac magnetic resonance (CMR) short-axis cine images to predict left ventricular ejection fraction (LVEF). Methods ·A total of 100 left ventricular hypertrophy (LVH) patients who visited the Department of Cardiology, Renji Hospital, Shanghai Jiao Tong University School of Medicine from January 2018 to December 2021, as well as 100 healthy control (HC) subjects during the same period, were included. All subjects completed CMR examinations under the supervision of experienced cardiologists and radiologists. The endocardial and epicardial contours were then manually delineated by cardiologists. Measurements of cardiac function and morphology were completed and data was recorded, including LVEF, left ventricular end-diastolic volume (LVEDV), and left ventricular end-diastolic mass (LVEDM). Myocardial 3D radiomic features of CMR-cine sequences were extracted by the Pyradiomics package, and selected and sorted by using correlation coefficient and K-best method. The LVEF prediction was performed with linear regression (LR), random forest (RF) and gradient boost (GB) methods. Results were also compared with LVEF prediction based on clinical information and CMR parameters. Results ·In terms of clinical indicators, there were significant differences between the LVH and HC groups, such as LVEDV and LVEDM (all P<0.05); after extracting 3D radiomic features, the top 10 features were selected for further analysis. LR regression model, GB regression model and RF regression model were constructed for predicting the LVEF, and RF regression models showed the best results with seven features, in which the mean absolute error (MAE) was 0.066±0.002. Further comparison results showed that the model using radiomic information with CMR parameters (MAE=0.056±0.001) had the best performance and it was significantly better than the model using radiomic features (MAE=0.066±0.002) or CMR parameters (MAE=0.060±0.001) alone (both P<0.05). Conclusion ·The use of radiomic features for LVEF prediction has certain feasibility, and combining radiomic features with CMR parameters can further improve the prediction accuracy of the model.

Key words: cardiac magnetic resonance (CMR), radiomics, left ventricular hypertrophy, cine sequence

CLC Number:

R542.2

LIU Qiming, LU Qifan, CHAI Yezi, JIANG Meng, PU Jun. Radiomics-based left ventricular ejection fraction prediction: a feasibility study[J]. Journal of Shanghai Jiao Tong University (Medical Science), 2023, 43(9): 1162-1168.

Figures/Tables 4

TrendMD

References 31

1	HUYNH K. Heart failure: improvement of LVEF in patients with HF is linked to better prognosis[J]. Nat Rev Cardiol, 2016, 13(9): 505.
2	LEWEY J, LEVINE L D, ELOVITZ M A, et al. Importance of early diagnosis in peripartum cardiomyopathy[J]. Hypertension, 2020, 75(1): 91-97.
3	Lang RM, Badano LP, Mor-Avi V, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging[J]. J Am Soc Echocardiogr, 2015, 28(1): 1-39.e14.
4	OMMEN S R, MITAL S, BURKE M A, et al. 2020 AHA/ACC guideline for the diagnosis and treatment of patients with hypertrophic cardiomyopathy: executive summary: a report of the American college of cardiology/American heart association joint committee on clinical practice guidelines[J]. Circulation, 2020, 142(25): e533-e557.
5	LAVIE C J, HARMON K G. Routine ECG screening of young athletes[J]. J Am Coll Cardiol, 2016, 68(7): 712-714.
6	SHIGA T, WAJIMA Z, INOUE T, et al. Survey of observer variation in transesophageal echocardiography: comparison of anesthesiology and cardiology literature[J]. J Cardiothorac Vasc Anesth, 2003, 17(4): 430-442.
7	IBANEZ B, ALETRAS A H, ARAI A E, et al. Cardiac MRI endpoints in myocardial infarction experimental and clinical trials[J]. J Am Coll Cardiol, 2019, 74(2): 238-256.
8	KWAN A C, POURMORTEZA A, STUTMAN D, et al. Next-generation hardware advances in CT: cardiac applications[J]. Radiology, 2021, 298(1): 3-17.
9	YU X, YAO X X, WU B F, et al. Using deep learning method to identify left ventricular hypertrophy on echocardiography[J].Int J Cardiovasc Imaging, 2022, 38(4): 759-769.
10	SALIBA L J, MAFFETT S. Hypertensive heart disease and obesity: a review[J]. Heart Fail Clin, 2019, 15(4): 509-517.
11	GEORGIOPOULOS G, FIGLIOZZI S, PATERAS K, et al. Comparison of demographic, clinical, biochemical, and imaging findings in hypertrophic cardiomyopathy prognosis: a network meta-analysis[J]. JACC Heart Fail, 2023, 11(1): 30-41.
12	FUNADA A, KANZAKI H, NOGUCHI T, et al. Prognostic significance of late gadolinium enhancement quantification in cardiac magnetic resonance imaging of hypertrophic cardiomyopathy with systolic dysfunction[J]. Heart Vessels, 2016, 31(5): 758-770.
13	LYU Q, SHAN H M, XIE Y B, et al. Cine cardiac MRI motion artifact reduction using a recurrent neural network[J]. IEEE Trans Med Imaging, 2021, 40(8): 2170-2181.
14	LARKMAN D J, HERLIHY A H, COUTTS G A, et al. Elimination of magnetic field foldover artifacts in MR images[J]. J Magn Reson Imaging, 2000, 12(5): 795-797.
15	RAJIAH P, KAY F, BOLEN M, et al. Cardiac magnetic resonance in patients with cardiac implantable electronic devices: challenges and solutions[J]. J Thorac Imaging, 2020, 35(1): W1-W17.
16	FERREIRA P F, GATEHOUSE P D, MOHIADDIN R H, et al. Cardiovascular magnetic resonance artefacts[J]. J Cardiovasc Magn Reson, 2013, 15(1): 41.
17	GILLIES R J, KINAHAN P E, HRICAK H. Radiomics: images are more than pictures, they are data[J]. Radiology, 2016, 278(2): 563-577.
18	LAMBIN P, LEIJENAAR R T H, DEIST T M, et al. Radiomics: the bridge between medical imaging and personalized medicine[J]. Nat Rev Clin Oncol, 2017, 14(12): 749-762.
19	KELLER H, WANGER K C, GOEPFRICH M, et al. Morphological quantification and differentiation of left ventricular hypertrophy in hypertrophic cardiomyopathy and hypertensive heart disease. A two dimensional echocardiographic study[J]. Eur Heart J, 1990, 11(1): 65-74.
20	ZWANENBURG A, VALLIÈRES M, ABDALAH M A, et al. The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping[J]. Radiology, 2020, 295(2): 328-338.
21	VAN GRIETHUYSEN J J M, FEDOROV A, PARMAR C, et al. Computational radiomics system to decode the radiographic phenotype[J]. Cancer Res, 2017, 77(21): e104-e107.
22	PEDREGOSA F, VAROQUAUX G, GRAMFORT A, et al. Scikit-learn: machine learning in python[EB/OL]. (2018-06-05) [2023-03-23]. https://arxiv.org/abs/1201.0490.
23	BUSTIN A, FUIN N, BOTNAR R M, et al. From compressed-sensing to artificial intelligence-based cardiac MRI reconstruction[J]. Front Cardiovasc Med, 2020, 7: 17.
24	VARRIANO G, GUERRIERO P, SANTONE A, et al. Explainability of radiomics through formal methods[J]. Comput Methods Programs Biomed, 2022, 220: 106824.
25	FAHMY A S, ROWIN E J, ARAFATI A, et al. Radiomics and deep learning for myocardial scar screening in hypertrophic cardiomyopathy[J]. J Cardiovasc Magn Reson, 2022, 24(1): 40.
26	RAISI-ESTABRAGH Z, JAGGI A, GKONTRA P, et al. Cardiac magnetic resonance radiomics reveal differential impact of sex, age, and vascular risk factors on cardiac structure and myocardial tissue[J]. Front Cardiovasc Med, 2021, 8: 763361.
27	WANG S, PATEL H, MILLER T, et al. AI based CMR assessment of biventricular function[J]. JACC Cardiovasc Imaging, 2022, 15(3): 413-427.
28	ZHENG X Y, YAO Z, HUANG Y N, et al. Deep learning radiomics can predict axillary lymph node status in early-stage breast cancer[J]. Nat Commun, 2020, 11(1): 1236.
29	BERNARD O, LALANDE A, ZOTTI C, et al. Deep learning techniques for automatic MRI cardiac multi-structures segmentation and diagnosis: is the problem solved?[J]. IEEE Trans Med Imaging, 2018, 37(11): 2514-2525.
30	CAMPELLO V M, MARTÍN-ISLA C, IZQUIERDO C, et al. Minimising multi-centre radiomics variability through image normalisation: a pilot study[J]. Sci Rep, 2022, 12(1): 12532.
31	PIANTADOSI G, SANSONE M, FUSCO R, et al. Multi-planar 3D breast segmentation in MRI via deep convolutional neural networks[J]. Artif Intell Med, 2020, 103: 101781.

Indicator	Total (n=200)	HC group (n=100)	LVH group (n=100)	P value
Demographic information
Age/year	53 (36, 62)	53 (36, 62)	53 (36, 63)	0.562
Male/n(%)	143 (71.5)	71 (71.0)	72 (72.0)	0.876
Weight/kg	72 (63, 80)	70±14	75 (65, 81)	0.014
Height/cm	170 (163, 175)	169±10	169±8	0.793
BMI/(kg·m^-2)	24.7 (22.3, 27.1)	24.2±3.5	26.0±4.0	0.000
BSA/m²	1.80±0.22	1.77±0.22	1.84 (1.70, 1.94)	0.051
CMR parameter
LVEF/%	65.1 (59.1, 70.2)	65.3±7.0	65.3 (55.2, 71.0)	0.336
LVEDV/mL	127.3 (107.1, 158.2)	125.7±34.7	134.8 (117.0, 167.8)	0.000
LVEDV/BSA/(mL·m^-2)	101.7 (80.1, 135.5)	70.2±14.2	75.9 (65.2, 91.8)	0.000
LVEDM/g	72.2 (62.7, 83.7)	83.7±24.0	134.1 (105.5, 175.6)	0.000
LVEDM/BSA/(g·m^-2)	55.5 (46.7, 71.9)	46.6±9.6	72.0 (59.5, 89.2)	0.000

Indicator	Total (n=200)	HC group (n=100)	LVH group (n=100)	P value
Demographic information
Age/year	53 (36, 62)	53 (36, 62)	53 (36, 63)	0.562
Male/n(%)	143 (71.5)	71 (71.0)	72 (72.0)	0.876
Weight/kg	72 (63, 80)	70±14	75 (65, 81)	0.014
Height/cm	170 (163, 175)	169±10	169±8	0.793
BMI/(kg·m^-2)	24.7 (22.3, 27.1)	24.2±3.5	26.0±4.0	0.000
BSA/m²	1.80±0.22	1.77±0.22	1.84 (1.70, 1.94)	0.051
CMR parameter
LVEF/%	65.1 (59.1, 70.2)	65.3±7.0	65.3 (55.2, 71.0)	0.336
LVEDV/mL	127.3 (107.1, 158.2)	125.7±34.7	134.8 (117.0, 167.8)	0.000
LVEDV/BSA/(mL·m^-2)	101.7 (80.1, 135.5)	70.2±14.2	75.9 (65.2, 91.8)	0.000
LVEDM/g	72.2 (62.7, 83.7)	83.7±24.0	134.1 (105.5, 175.6)	0.000
LVEDM/BSA/(g·m^-2)	55.5 (46.7, 71.9)	46.6±9.6	72.0 (59.5, 89.2)	0.000

Model	MAE	P value^①
LR	0.075±0.000	0.000
GB	0.072±0.000	0.000
RF	0.066±0.002	-

Model	MAE	P value^①
LR	0.075±0.000	0.000
GB	0.072±0.000	0.000
RF	0.066±0.002	-

Feature group^①	MAE	P value^②
Basic information^③	0.071±0.002	0.000
Radiomics	0.066±0.002	0.000
CMR^④	0.060±0.001	0.001
CMR+radiomics	0.056±0.001	NA