Fit of biokinetic data in molecular radiotherapy: a machine learning approach.
Akaike information criterion
Biokinetic curves
F-test
Fit function
Machine learning
Journal
EJNMMI physics
ISSN: 2197-7364
Titre abrégé: EJNMMI Phys
Pays: Germany
ID NLM: 101658952
Informations de publication
Date de publication:
22 Feb 2024
22 Feb 2024
Historique:
received:
13
06
2023
accepted:
15
02
2024
medline:
22
2
2024
pubmed:
22
2
2024
entrez:
21
2
2024
Statut:
epublish
Résumé
In literature are reported different analytical methods (AM) to choose the proper fit model and to fit data of the time-activity curve (TAC). On the other hand, Machine Learning algorithms (ML) are increasingly used for both classification and regression tasks. The aim of this work was to investigate the possibility of employing ML both to classify the most appropriate fit model and to predict the area under the curve (τ). Two different ML systems have been developed for classifying the fit model and to predict the biokinetic parameters. The two systems were trained and tested with synthetic TACs simulating a whole-body Fraction Injected Activity for patients affected by metastatic Differentiated Thyroid Carcinoma, administered with [ As N varies, CA remains constant for ML (about 98%), while it improves for F-test (from 62 to 92%) and AICc (from 50 to 92%), as N increases. With AM, [Formula: see text] can reach down to - 67%, while using ML [Formula: see text] ranges within ± 25%. Using real TACs, there is a good agreement between τ obtained with ML system and AM. The employing of ML systems may be feasible, having both a better classification and a better estimation of biokinetic parameters.
Sections du résumé
BACKGROUND
BACKGROUND
In literature are reported different analytical methods (AM) to choose the proper fit model and to fit data of the time-activity curve (TAC). On the other hand, Machine Learning algorithms (ML) are increasingly used for both classification and regression tasks. The aim of this work was to investigate the possibility of employing ML both to classify the most appropriate fit model and to predict the area under the curve (τ).
METHODS
METHODS
Two different ML systems have been developed for classifying the fit model and to predict the biokinetic parameters. The two systems were trained and tested with synthetic TACs simulating a whole-body Fraction Injected Activity for patients affected by metastatic Differentiated Thyroid Carcinoma, administered with [
RESULTS
RESULTS
As N varies, CA remains constant for ML (about 98%), while it improves for F-test (from 62 to 92%) and AICc (from 50 to 92%), as N increases. With AM, [Formula: see text] can reach down to - 67%, while using ML [Formula: see text] ranges within ± 25%. Using real TACs, there is a good agreement between τ obtained with ML system and AM.
CONCLUSIONS
CONCLUSIONS
The employing of ML systems may be feasible, having both a better classification and a better estimation of biokinetic parameters.
Identifiants
pubmed: 38383799
doi: 10.1186/s40658-024-00623-5
pii: 10.1186/s40658-024-00623-5
doi:
Types de publication
Journal Article
Langues
eng
Pagination
19Informations de copyright
© 2024. The Author(s).
Références
Dewaraja YK, Schipper MJ, Shen J, et al. Tumor-absorbed dose predicts progression-free survival following (131)I-tositumomab radioimmunotherapy. J Nucl Med. 2014;55(7):1047–53. https://doi.org/10.2967/jnumed.113.136044 .
doi: 10.2967/jnumed.113.136044
pubmed: 24842891
Bolch WE, Eckerman KF, Sgouros G, et al. MIRD pamphlet No. 21: a generalized schema for radiopharmaceutical dosimetry-standardization of nomenclature. J Nucl Med. 2009;50(3):477–84. https://doi.org/10.2967/jnumed.108.056036 .
doi: 10.2967/jnumed.108.056036
pubmed: 19258258
Siegel JA, Thomas SR, Stubbs JB, et al. MIRD pamphlet no. 16: Techniques for quantitative radiopharmaceutical biodistribution data acquisition and analysis for use in human radiation dose estimates. J Nucl Med. 1999;40(2):37S-61S.
pubmed: 10025848
Glatting G, Kletting P, Reske SN, et al. Choosing the optimal fit function: comparison of the Akaike information criterion and the F-test. Med Phys. 2007;34(11):4285–92. https://doi.org/10.1118/1.2794176 .
doi: 10.1118/1.2794176
pubmed: 18072493
Akaike H. Information theory and an extension of the maximum likelihood principle. In: Petrov BN, Csaki F, editors. Second international symposium on inference theory. Budapest: Akademiai Kiado; 1973. p. 267–81.
Burnham KP, Anderson DR. Model selection and multimodel interference-a practical information-theoretic approach. New York: Springer; 2002.
Motulsky HJ, Christopoulos A. Fitting models to biological data using linear and nonlinear regression: a practical guide to curve fitting. Boston: GraphPad Software Inc; 2004.
doi: 10.1093/oso/9780195171792.001.0001
Hurvich CM, Tsai CL. Regression and time series model selection in small samples. Biometrika. 1989;76(2):297–307. https://doi.org/10.1093/biomet/76.2.297 .
doi: 10.1093/biomet/76.2.297
Kletting P, Schimmel S, Kestler HA, et al. Molecular radiotherapy: the NUKFIT software for calculating the time-integrated activity coefficient. Med Phys. 2013;40(10):102504. https://doi.org/10.1118/1.4820367 .
doi: 10.1118/1.4820367
pubmed: 24089925
Currie G, Hawk KE, Rohren E, et al. Machine learning and deep learning in medical imaging: intelligent imaging. J Med Imaging Radiat Sci. 2019;50(4):477–87. https://doi.org/10.1016/j.jmir.2019.09.005 .
doi: 10.1016/j.jmir.2019.09.005
pubmed: 31601480
Uribe CF, Mathotaarachchi S, Gaudet V, et al. Machine learning in nuclear medicine: part 1-introduction. J Nucl Med. 2019;60(4):451–8. https://doi.org/10.2967/jnumed.118.223495 .
doi: 10.2967/jnumed.118.223495
pubmed: 30733322
Zukotynski K, Gaudet V, Uribe CF, et al. Machine learning in nuclear medicine: part 2-neural networks and clinical aspects. J Nucl Med. 2021;62(1):22–9. https://doi.org/10.2967/jnumed.119.231837 .
doi: 10.2967/jnumed.119.231837
pubmed: 32978286
Flux GD, Guy MJ, Beddows R, Pryor M, et al. Estimation and implications of random errors in whole-body dosimetry for targeted radionuclide therapy. Phys Med Biol. 2002;47(17):3211–23. https://doi.org/10.1088/0031-9155/47/17/311 .
doi: 10.1088/0031-9155/47/17/311
pubmed: 12361219
Géron A. Hands-on machine learning with scikit-learn, keras, and TensorFlow. Sebastopol: O’Reilly Media; 2019.
Lassmann M, Hänscheid H, Chiesa C, et al. EANM Dosimetry Committee. EANM Dosimetry Committee series on standard operational procedures for pre-therapeutic dosimetry I: blood and bone marrow dosimetry in differentiated thyroid cancer therapy. Eur J Nucl Med Mol Imaging. 2008;35(7):1405–12. https://doi.org/10.1007/s00259-008-0761-x .
doi: 10.1007/s00259-008-0761-x
pubmed: 18491092
Schapire RE. The strength of weak learnability. Mach Learn. 1990;5:197–227. https://doi.org/10.1007/BF00116037 .
doi: 10.1007/BF00116037
https://scikit-learn.org/
Kirkup L, Sutherland J. Curve stripping and nonlinear fitting of polyexponential functions to data using a microcomputer. Comput Phys. 1988;2:64. https://doi.org/10.1063/1.168313 .
doi: 10.1063/1.168313
Kletting P, Glatting G. Model selection for time-activity curves: the corrected Akaike information criterion and the F-test. Z Med Phys. 2009;19(3):200–6. https://doi.org/10.1016/j.zemedi.2009.05.003 .
doi: 10.1016/j.zemedi.2009.05.003
pubmed: 19761098
Sarrut D, Halty A, Badel JN, et al. Voxel-based multimodel fitting method for modeling time activity curves in SPECT images. Med Phys. 2017;44(12):6280–8. https://doi.org/10.1002/mp.12586 .
doi: 10.1002/mp.12586
pubmed: 28921613