Post-doctoral position on Radiomics and Random Forest Survival.
Applications on Multiple Myeloma and lymphoma malignancies
(mantle cell and diffuse large B-cell) explored by PET/CT
Host institutions:
• University Hospital of Nantes, Nuclear Medicine Department / French Institute of Health and Medical Research (INSERM, CRCINA, Nuclear Oncology Team, UMR1232), Nantes, France
• Ecole Centrale Nantes, LS2N, CNRS UMR 6004, Nantes, France
Supervision:
• Dr Thomas Carlier (thomas.carlier@chu-nantes.fr)
• Pr Diana Mateus (diana.mateus@ec-nantes.fr)
Duration: 2 years
Context and objectives: Personalized medicine is one of the major goals of oncology. With the help of technological breakthroughs, the fine characterization of tumors has led to the identification of diagnostic biomarkers, which are prognostic of survival or predictive of the response to therapeutic agents. PET/CT imaging is an integral part of this approach by enabling the distribution and accessibility of biomarkers expressed by tumor or microenvironment in vivo in a non-invasive way. To date, numerous studies explored the potential value of textural features in PET imaging with encouraging results in a number of cancers [1, 2]. However, only a limited number adopted rigorous methodological choices with in particular large cohorts of patients and robust statistical analysis [3, 4]. Keeping in mind these limitations, the evidence supporting the additional value of advanced image features from FDG-PET continues to expand year after year. Several of the most recent studies have used techniques such as external cohort validation [5, 6], and machine(deep)-learning technique [7, 8] and concluded in the usefulness of textural analysis regarding patient management.
Our team developed an approach based on Random Survival Forest [9] in the context of patient suffering from multiple myeloma using PET imaging at baseline. The aim of this post-doctoral position will be to enhance this methodology taking into account unbalanced data, right censoring, boosting… and to assess the benefits of fractal analysis as an alternative textural features. A second aim will be to improve the interpretability of RSFs by studying the importance of the most predictive features for very high dimensional and correlated data [10, 11]. The candidate will apply the development within the context of several large multicentric, prospective studies including IMAJEM fo MM patients [12], LYMA for mantle cell lymphoma [13] and GAINED for DLBCL (https://clinicaltrials.gov/ct2/ show/NCT01659099).
This position is funded by a large project called SIRIC (ILIAD Imaging and Longitudinal Investigations to Ameliorate Decision making in multiple myeloma and breast cancer) involving several teams (from biology to applied mathematics) in Nantes. This project is conducted in strong partnership with the Numerical Science Laboratory of Nantes and will be associated with the work done by a current PhD student.
Requirements:
• Education: The candidate must hold a PhD in Physics, Computer Science or Applied Mathematics
• Programming Skills: Python, R
References
[1] M. Hatt, F. Tixier, L. Pierce, P. E. Kinahan, C. C. Le Rest, and D. Visvikis, “Characterization of PET/CT images using texture analysis: the past, the present… any future?” Eur J Nucl Med Mol Imaging, vol. 44, no. 1, pp. 151–165, Jan 2017.
[2] J. W. Lee and S. M. Lee, “Radiomics in Oncological PET/CT: Clinical Applications,” Nucl Med Mol Imaging, vol. 52, no. 3, pp. 170–189, Jun 2018.
[3] A. Chalkidou, M. J. O’Doherty, and P. K. Marsden, “False Discovery Rates in PET and CT Studies with Texture Features: A Systematic Review,” PLoS ONE, vol. 10, no. 5, p. e0124165, 2015.
[4] M. L. Welch, C. McIntosh, B. Haibe-Kains, M. F. Milosevic, L. Wee, A. Dekker, S. H. Huang, T. G. Purdie, B. O’Sullivan, H. J. W. L. Aerts, and D. A. Jaffray, “Vulnerabilities of radiomic signature development: The need for safeguards,” Radiother Oncol, vol. 130, pp. 2–9, 01 2019.
[5] F. Lucia, D. Visvikis, M. Vallieres, M. C. Desseroit, O. Miranda, P. Robin, P. A. Bonaffini, J. Alfieri, I. Masson, A. Mervoyer, C. Reinhold, O. Pradier, M. Hatt, and U. Schick, “External validation of a combined PET and MRI radiomics model for prediction of recurrence in cervical cancer patients treated with chemoradiotherapy,” Eur J Nucl Med Mol Imaging, vol. 46, no. 4, pp. 864–877, Apr 2019.
[6] S. Carvalho, R. T. H. Leijenaar, E. G. C. Troost, J. E. van Timmeren, C. Oberije, W. van Elmpt, L. F. de Geus- Oei, J. Bussink, and P. Lambin, “18F-fluorodeoxyglucose positron-emission tomography (FDG-PET)-Radiomics of metastatic lymph nodes and primary tumor in non-small cell lung cancer (NSCLC) – A prospective externally validated study,” PLoS ONE, vol. 13, no. 3, p. e0192859, 2018.
[7] P. P. Ypsilantis, M. Siddique, H. M. Sohn, A. Davies, G. Cook, V. Goh, and G. Montana, “Predicting Response to Neoadjuvant Chemotherapy with PET Imaging Using Convolutional Neural Networks,” PLoS ONE, vol. 10, no. 9, p.e0137036, 2015.
[8] H. Arimura, M. Soufi, H. Kamezawa, K. Ninomiya, and M. Yamada, “Radiomics with artificial intelligence for precision medicine in radiation therapy,” J Radiat Res, vol. 60, no. 1, pp. 150–157, Jan 2019.
[9] L. Morvan, T. Carlier, B. Jamet, C. Bailly, C. Bodet-Milin, P. Moreau, F. Kraeber-Bodere, and D. Mateus, “Leveraging RSF and PET images for prognosis of multiple myeloma at diagnosis,” Int J Comput Assist Radiol Surg, Jun 2019.
[10] R. Genuer, J. Poggi, C. Tuleau-Malot, and N. Villa-Vialaneix, “Random Forests for big data,” Big Data Res, vol. 9, pp. 28–46, 2017.
[11] B. Gregorutti, B. Michel, and P. Saint-Pierre, “Correlation and variable importance in random forests,” Stat Comput, vol. 27, p. 659, 2017.
[12] P. Moreau, M. Attal, D. Caillot, M. Macro, L. Karlin, L. Garderet, T. Facon, L. Benboubker, M. Escoffre-Barbe, A. M. Stoppa, K. Laribi, C. Hulin, A. Perrot, G. Marit, J. R. Eveillard, F. Caillon, C. Bodet-Milin, B. Pegourie, V. Dorvaux, C. Chaleteix, K. Anderson, P. Richardson, N. C. Munshi, H. Avet-Loiseau, A. Gaultier, J. M. Nguyen, B. Dupas, E. Frampas, and F. Kraeber-Bodere, “Prospective Evaluation of Magnetic Resonance Imaging and [18F]Fluorodeoxyglucose Positron Emission Tomography-Computed Tomography at Diagnosis and Before Main-
tenance Therapy in Symptomatic Patients With Multiple Myeloma Included in the IFM/DFCI 2009 Trial: Results of the IMAJEM Study,” J Clin Oncol, vol. 35, no. 25, pp. 2911–2918, Sep 2017.
[13] C. Bailly, T. Carlier, A. Berriolo-Riedinger, O. Casasnovas, E. Gyan, M. Meignan, A. Moreau, B. Burroni, L. Djaileb, R. Gressin, A. Devillers, T. Lamy, C. Thieblemont, O. Hermine, F. Kraeber-Bodere, S. Le Gouill, and C. Bodet- Milin, “Prognostic value of FDG-PET in patients with mantle cell lymphoma: results from the LyMa-PET Project,”Haematologica, Aug 2019.