Radiomics- and dosiomics-based prediction of treatment failure after chemoradiotherapy for pharyngeal squamous cell carcinoma patients

doi:10.21203/rs.3.rs-3588937/v1

Download PDF

Research Article

Radiomics- and dosiomics-based prediction of treatment failure after chemoradiotherapy for pharyngeal squamous cell carcinoma patients

https://doi.org/10.21203/rs.3.rs-3588937/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

We investigated an approach to predict treatment failure after chemoradiation using radiomics and dosiomics in patients with pharyngeal squamous cell carcinoma (PSCC). Data of 172 patients were collected from a public database. The cases were divided into the training (n = 140) and testing (n = 32) datasets. A total of 1027 features, including radiomic (R) features (first-order, texture, and wavelet features), local binary pattern-based (L) features, and topological (T) features, were extracted from the computed tomography (CT) image and dose distribution (DD) of the gross tumor volume. The Coxnet algorithm was employed on the training dataset to select significant features. Twenty-One treatment failure prediction models were constructed based on Rad scores. The overall adequacy of the treatment failure prediction models was evaluated using the concordance index (C-index) and statistically significant differences (p-values) between the Kaplan–Meier curves of the two risk groups. The DD-based LT (DD-LT) model and the combined CT with DD-based RLT (CD-RLT) model showed statistically significant differences in the Kaplan–Meier curves. The C-indices were 0.74 for the DD-LT model and 0.64, the CD-RLT model. These models exhibited higher performance than the conventional approach. The proposed radiomics/dosiomics-based model may be more accurate in predicting treatment failure after chemoradiation in patients with PSCC.

treatment failure prediction

pharyngeal squamous cell carcinoma

topology

radiomics

dosiomics

Pharyngeal squamous cell carcinoma (PSCC) affects approximately 136,000 people annually worldwide [1]. The 5-year overall survival rate (5y-OS) for PSCC is estimated to be approximately 60%. Furthermore, advanced PSCCs (stages III and IV) have a poor prognosis, with 5y-OS of < 50%. Radiotherapy (RT) combined with chemotherapy (chemoradiotherapy [CRT]) plays an important role in the definitive treatment of PSCC. However, the 5-year survival rate remains low despite advances in treatment owing to treatment failure. Failure after PSCC treatment of PSCCs include recurrence (local recurrence, locoregional recurrence, and regional recurrence), distant metastasis, and residual tumors. Recurrence occurs in 16–52% of patients with head and neck squamous cell carcinoma (HNSCC) treated with definitive CRT [2–7]. Therefore, the prediction of treatment failure in each patient with PSCC prior to treatment planning can lead to more optimal treatment plans or select a better treatment approach.

In recent years, radiomics and dosiomics have been widely applied for recurrence prediction of HNSCC [8–10], preoperative prediction of the malignancy of parotid gland tumors [11, 12], and prognosis prediction in patients with HNSCC [13]. Radiomics is a rapidly developing field that analyzes the relationships between clinical outcomes and features extracted from medical images. In addition, dosiomics is a radiomics approach, and the dose distribution simulated in RT treatment planning is analyzed instead of using medical images. Dose distribution is closely related to tumor control in RT. Tumors consist of heterogeneous tumor cells as well as resistant and sensitive cells. Therefore, inadequate dose distribution of an insufficient dose to resistant cells can cause treatment failure. We have previously reported on recurrence prediction after RT in patients with HNSCC using dosiomics [10]. This performance was better than that of conventional radiomics approaches.

Keek et al. reported two CT-based radiomics models constructed with a random survival forest to predict locoregional recurrence (LRR) and distant metastasis (DM) after concurrent CRT in patients with HNSCC [8]. Both models were constructed using a random survival forest. The concordance indices (C-indices) as prediction performance indicator of radiomics-LRR and radiomics-DM models were 0.52 and 0.49, respectively. Wu et al. presented models for LRR recurrence prediction of HNSCC using radiomics-based, dosiomics-based, and combined approaches [9]. The C-indices were 0.54 for the computed tomography (CT) image-based radiomics, 0.60 for the dosiomics, and 0.60 for the mixed CT-based radiomics and dosiomics models. Moreover, Kamezawa et al. reported recurrence prediction after RT for patients with HNSCC using dosiomics applied with the local binary pattern. The prediction performance showed a C-index of 0.76 [10]. However, most previous studies analyzed mixed data from various tumor sites, including the oral cavity, tongue, tonsils, pharynx, and larynx. Therefore, in this study, we focused on the prediction of treatment failure for PSCC.

Topological radiomics is a novel approach in radiomics research [12, 13]. It is analyzed using topological features, particularly Betti numbers, extracted from medical images. Betti numbers are topological invariants used in algebraic topology to characterize the connectivity and shapes of spaces. Betti numbers provide information regarding the number of connected components, holes, and higher-dimensional cavities in the data [14–16]. In radiomics, the Betti numbers can be used to analyze and quantify the topological properties of medical images. Higher Betti values indicate a greater number of holes or voids in the image, which can be associated with different pathological conditions or anatomical structures. In dosiomics, topological features may represent nonhomogeneous dose distributions related to treatment failure. Therefore, topological features based on the Betti number may reveal the tumor characteristics. They may improve the predictive performance of tumor characterization, prognosis, and treatment response. Ikushima et al. reported topological radiomics for predicting the malignancy of parotid gland tumors [12]. The topology-based approach showed better performance (accuracy: 0.975) than the conventional approach (accuracy: 0.795). Le et al. suggested that topology-based radiomics can capture intrinsic tumor information and predict the prognosis of patients with HNSCC [13]. The C-index for prediction performance was 0.80. Therefore, a topology-based approach may be useful in predicting treatment failure in patients with PSCC.

To the best of our knowledge, no studies have considered the application of topology to radiomics- and dosiomics-based analyses of HNSCC and PSCC. Therefore, we propose a novel approach for radiomics- and dosiomics-based analyses that includes topological features to improve the accuracy of treatment failure prediction in patients with PSCC after CRT.

2.1 Overview

Figure 1 provides an overview of this study. First, the gross tumor volume (GTV) region on the CT image and dose distribution (DD) in the GTV region were cropped based on the GTV contour information. Radiomics (R), local binary pattern (L), and topological (T) features were extracted from the cropped CT and DD images. A total of 21 models, that is R, L, T, RL, RT, LT, and RLT in CT only, DD only, and combined CT with DD (CD), were compared. Second, significant features associated with treatment failure were explored using the Cox proportional hazards (CPH) model with the least absolute shrinkage and selection operator (LASSO) penalty. The treatment failure prediction model was then constructed with the Rad-score calculated using the significant features. Finally, the performance of the prediction model was assessed using the C-index and Kaplan–Meier curves.

2.2 Patients

Overall, 172 cases, which were treated with CRT to the PSCC of stage III and IV were collected from the “HNSCC” and “HN-PET-CT” database in The Cancer Imaging Archive (TCIA) [17–22]. Institutional Review Board approval was not required because TCIA is an open-access database and all data collected for this study were freely available for scientific research. Moreover, our study followed the TCIA data-usage policy. The cases were divided into the training (n = 140) and testing (n = 32) datasets. Patient characteristics are shown in Table 1. Local recurrence, locoregional recurrence, regional recurrence, distant metastasis, and residual tumors were considered treatment failures after CRT. The numbers of treatment failures were 28 and 16 in the training and test datasets, respectively. The number of treatment failures and complete response (CR) cases were imbalanced at 28 and 112, respectively. Therefore, treatment failure in the training dataset was balanced using the synthetic minority oversampling technique (SMOTE) [23], which can randomly generate synthetic data from a minority group, while maintaining the distribution of the original data. Therefore, the number of treatment failure cases in the training dataset increased to 112, which is the same as the number of CR cases.

Table 1

Patient characteristics.
	Training	Test	p-value
Total number of cases	140	32
Age [y, min-max (median)]	29–91 (62)	49–82 (64)	0.23^*
Sex			0.28⁺
Male	111	24
Female	29	8
Stage			0.75⁺
III	18	6
IV	122	26
Tumor site			0.50⁺
Nasopharynx	5	2
Oropharynx	125	28
Hypopharynx	10	2
Tumor status after CRT			1.00⁺
Treatment failure	28 [112^†]	16
Complete response	112	16
Treatment failure-free survival time [y, min-max (median)]	0.39–8.64 (3.80)	0.46–8.62 (3.08)	0.08^*
CRT: chemoradiotherapy, ^† Number of augmented data, * Mann-Whitney U-test, + Fisher's exact test

2.3 Feature extraction

Prior to feature extraction, the CT images and DD in the GTV region were cropped based on the contour information of the GTV. Preprocessing, which resamples the voxel size and quantization, for CT and DD was then performed. The CT and DD images were resampled with an isotropic voxel size of 1 × 1 × 1 mm³. Quantization of CT and DD was performed using 64 bits and a fixed bin width of 25 cGy.

In total, 486 R, 54 L, and 309 T features were extracted from the CT images, and 55 R, 54 L, and 69 T features were extracted from the DD images. All the features were extracted using an in-house program implemented in MATLAB R2022b (MathWorks, Natick, MA, USA). The extracted features are listed in Table 2. Ultimately, each feature was standardized using a Z-score.

Table 2

Extracted features.
Category	Data	Feature	No. of features
R	CT	First-order features	14
		Texture features (GLCM, GLRLM, GLSZM, and NGTDM)	40
		Wavelet decomposition features	432
	DD	First-order features	14
		Texture features (GLCM, GLRLM, GLSZM, and NGTDM)	40
		Cold spot volume	1
L	CT	First-order features	14
	CT	Texture features (GLCM, GLRLM, GLSZM, and NGTDM)	40
	DD	First-order features	14
	DD	Texture features (GLCM, GLRLM, GLSZM, and NGTDM)	40
T	CT	Betti number (b0, b2, and b2/b0)	30
		Maximum of Betti number	3
		Area under the Betti number curve	3
	DD	Betti number (b0, b2, and b2/b0)	63
		Maximum of Betti number	3
		Area under the Betti number curve	3
		Total	1027
R: radiomic feature, L: local binary pattern feature, T: topological feature, CT: computed tomography, DD: dose distribution, GLCM: gray-level co-occurrence matrix, GLRLM: gray-level run length matrix, GLSZM: gray-level size zone matrix, NGTDM: neighborhood gray tone matrix

2.3.1 Radiomics (R) feature

The R feature group consisted of first-order (n = 14), texture (n = 40), and wavelet decomposition (WD) features (n = 432). However, the WD features were calculated from CT only, and the cold-spot volume (n = 1) was added only to the DD. Cold spots were defined as 95% of the prescribed dose, according to the Radiation Therapy Oncology Group 1016 protocol [24]. Therefore, R values for CT and DD (CT-R and DD-R) were 486 and 55, respectively. Texture features were calculated from four categories: the gray-level cooccurrence matrix (GLCM), gray-level run length matrix (GLRLM), gray-level size zone matrix (GLSZM), and neighborhood gray-tone matrix (NGTDM [25–28]. WD was performed by applying either a low- (L) or high-pass filter (H) in each direction of the three-dimensional volume using a Coiflet1 mother wavelet [29]. The 432 WD features were derived by calculating 14 first-order and 40 texture features for each of the eight wavelet decomposition images: LLL, LLH, LHL, LHH, HLL, HLH, HHL, and HHH.

2.3.2 Local binary pattern (L) feature

Local binary pattern (L) was transformed from the original CT and DD images using the following equation [30]:

$$L=\sum _{p=1}^{PS-1}s\left({v}_{p}-{v}_{c}\right){2}^{p}$$

where $p$ is the position number of a neighboring pixel, $PS$ is the patch size, ${v}_{c}$ is the central pixel value in an arbitrary patch (as a threshold value), ${v}_{p}$ is the neighboring pixel value of position $p$ in an arbitrary patch, and.

$$\text{s}\left(x\right)=\left\{\begin{array}{c}1, x\ge 0\\ 0, x<0\end{array}\right.$$

A $PS$ of 3 × 3 was employed for this study such that the number of neighboring pixels was eight ($p$:1–8). Fourteen first-order and 40 texture features were extracted from the CT-L and DD-L.

2.3.3 Topological (T) feature

Betti numbers are invariant for topological spaces based on connectivity. In three-dimensional image, two values, b₀ and b₂ were included for the Betti numbers. b₀ represents the number of connected components and b₂ is the number of cavities. b₀ and b₂ were calculated from binarized CT and DD by changing the threshold of CT value (range: mean CT value in CT ± 50, increment: 1 Hounsfield unit [HU]) and percentage prescribed dose (range: 95–105%, increment: 0.5%). In other words, b₀ and b₂ were obtained from each CT, and the DD was binarized with various thresholds. Moreover. The ratio of b₀ and b₂ (b₂/b₀) and area under the Betti number curve (AUBC) were employed as T features. AUBC is a definite integral of a curve describing the change in b₀, b₂, and b₂/b₀ as a function of the threshold value. An example of the Betti number curves in the CT of the four cases is shown in Fig. 2. The Betti number curves varied depending on each case and may be able to express the differences in cancer traits.

2.4 Prediction model construction and performance evaluation

Twenty-One models, that is, R, L, T, RL, RT, LT, and RLT models in CT-only, DD-only, and CD were constructed to compare treatment failure prediction performances. Significant features were selected using Coxnet, and the CPH model was optimized using the least absolute LASSO (L1 norm) penalty. LASSO is widely used in radiomics to select features. If the coefficients of the features are zero, they are considered irrelevant to treatment failure. In other words, only features with coefficients were related to treatment failure, and only significant features could be selected. In this study, the number of selected features was determined to satisfy the condition “N > 10 × n” (N: sample size;, n: number of selected features) [31]. The Coxnet algorithm using the “glmnet” package in R software (R Foundation for Statistical Computing, Vienna, Austria) version 4.0.2 was employed to select significant features in the training dataset. The coefficients of the features selected by Coxnet represent the degree of contribution of each feature to the prediction of treatment failure. Moreover, to consider multicollinearity, pairwise correlation coefficients between selected features were calculated, and no highly correlated features (> 0.70) were identified. The Rad-score was calculated as follows:

$$Rad-score=\sum _{i=1}^{n}{\beta }_{i}\times {SF}_{i}$$

where $n$ is the number of selected features, $\beta$ is the coefficient of the selected feature obtained by the Coxnet, and the $SF$ is the value of selected feature. Ultimately, a prediction model was constructed using a Rad-score-based CPH model.

C-indices and balanced C-indices were employed to assess model performance. The balanced C-index was calculated as the average value of the C-indices in the training and testing datasets. Moreover, all PSCC cases in each dataset were stratified into two groups by the median of the Rad-score, and the significant difference between the two groups was evaluated by Kaplan–Meier curves (treatment failure-free probability curves) with the log-rank test (p < 0.05).

Twenty-Two significant features associated with treatment failure were selected in the order of larger absolute values of the coefficients for the training dataset of 224. The number of selected features was determined as the minimum number of features selected from each of the 21 models. Figure 3 shows an example of CT and binary images of nontreatment failure (upper) and treatment failure cases (lower). In the binary images used to extract the T features, the treatment failure case had more cavities (b2), whereas the nontreatment failure case had almost a connected component (b0).

Table 3 shows the C-indices and p-values of the treatment failure-free probability curves of the high- and low-risk groups, stratified based on the median Rad-score in the treatment failure prediction model. The CT-R (conventional CT-radiomics) model had the highest balanced C-index of 0.76. However, there was no statistically significant difference in the treatment failure-free probability curves in the test dataset (p = 0.08). Statistically significant differences in the treatment failure-free probability curves appeared only in the DD-LT and CD-RLT models. Figure 4 shows the treatment failure-free probability curves for the DD-LT and CD-RLT models. In the training and test dataset, the statistically significant differences were observed as p = 0.005 and p = 0.02 for the DD-LT and p < 0.001 and p = 0.03 CD-RLT models. The C-indices of training and test datasets were 0.65 and 0.74 for the DD-LT model and 0.83 and 0.64 for the CD-RLT model.

Table 3

Performances of treatment failure prediction models.
	C-index		p-value
Model	Training	Test	Training	Test
CT-R	0.88	0.56	< 0.001	0.86
CT-L	0.84	0.50	< 0.001	0.81
CT-T	0.76	0.58	< 0.001	0.52
CT-RL	0.90	0.57	< 0.001	0.48
CT-RT	0.86	0.58	< 0.001	0.59
CT-LT	0.87	0.52	< 0.001	0.72
CT-RLT	0.93	0.53	< 0.001	0.70
DD-R	0.79	0.52	< 0.001	0.28
DD-L	0.67	0.64	< 0.001	0.29
DD-T	0.72	0.70	< 0.001	0.38
DD-RL	0.80	0.43	< 0.001	0.73
DD-RT	0.77	0.69	< 0.001	0.20
DD-LT	0.65	0.74	< 0.005	0.02
DD-RLT	0.80	0.60	< 0.001	0.73
CD-R	0.91	0.60	< 0.001	0.08
CD-L	0.80	0.52	< 0.001	0.80
CD-T	0.81	0.69	< 0.001	0.07
CD-RL	0.85	0.54	< 0.001	0.98
CD-RT	0.89	0.52	< 0.001	0.76
CD-LT	0.86	0.61	< 0.001	0.78
CD-RLT	0.83	0.64	< 0.001	0.03
CT: computed tomography, DD: dose distribution, CD: combined CT with DD

In this study, a novel approach for radiomics and dosiomics analysis was proposed. The treatment failure prediction model was able to stratify patients into low- and high-risk groups, the DD-LT and CD-RLT models may play a role in predicting treatment failure after CRT in patients with PSCC.

Keek et al. developed two radiomic-based models to predict LRR and DM after CRT in patients with HNSCC [8]. C indices for prediction performance were 0.52 for radiomics-LRR and 0.49 for the radiomics-DM models. Our topology-based DD-LT and CD-RLT models demonstrated the C-indices of 0.74 and 0.64, respectively. Therefore, our proposed novel approach can accurately predict treatment failure prior to treatment planning in patients with PSCC.

Wu et al. developed recurrence prediction models for patients with HNSCC [9]. The C-index of the conventional CT radiomics model was 0.54. The C-indices of the dosiomics and combined with the CT radiomics and dosiomics models were 0.60 and 0.60, respectively. By comparing our treatment failure prediction models (DD-LT and CD-RLT models) with their recurrence prediction models, our treatment failure models achieved a higher C-indices (0.74 and 0.64) than their models. Therefore, the treatment failure prediction performance after CRT in patients with PSCC may be improved by applying the topology or local binary pattern to conventional radiomics/dosiomics models. However, they reported that the CT radiomics/dosiomics prediction model combined with PET data showed improvement, with a C-index of 0.66. Hence, our model can potentially be improved by applying radiomics features extracted from PET images. Therefore, incorporation of PET images should be investigated in future studies.

LR, LRR, DM, and residual tumor were defined as treatment failures in this study owing to the insufficient number of cases for each outcome. This study aimed to predict treatment failure in patients with LR, LRR, DM, or residual tumors. Treatment failure is primarily caused by tumor resistance. Resistant tumor cells can survive treatment (residual tumor), contribute to LR and LRR, or lead to the development of DM. Thus, our approach may be able to extract resistant tumor cells from the phenotype in medical images through a more detailed analysis in the future.

This study had some limitations. First, interobserver variability in tumor delineation was not considered. This may affect the robustness and generalizability of the radiomics-based prediction models. Pavic et al. reported that the stability rate of radiomic features with respect to inter-observer delineation variability was 59% in HNSCC [32]. This could be due to the inclusion of several tumor sites. Therefore, it is necessary to limit the number of tumor sites used in the analysis as much as possible. Moreover, employing consensus-based approaches, such as obtaining agreement among multiple observers or expert consensus, may help mitigate inter-observer delineation variability. In addition, automated or semiautomated delineation techniques may address the problem of interobserver variability.

Second, treatment planning can vary by facility, planners (i.e., oncologists and medical physicists), years of experience, and dose calculation algorithms, leading to differences in the calculated dose distributions. This variability can affect the reliability and reproducibility of the dosiomic analyses. Bufacchi et al. calculated the target dose for nasopharyngeal carcinomas with the two dose-calculation algorithms [33]. As a result, the difference in tumor control probability reached 6.8%. Therefore, it is necessary to investigate the reliability and reproducibility of the dosiomic analysis of the prediction models in future studies.

Third, the number of selected features was not optimized for the construction of treatment failure prediction models. The optimal number of features for each model can differ. Therefore, there is scope for further study on the optimization of the number of selected features.

Finally, the treatment failure prediction model developed in this study is based on a limited number of patients with PSCC (n = 224). Therefore, it may be crucial to ensure an appropriate number of patients in order to improve the performance of treatment failure prediction models. Thus, a longitudinal study with a larger sample size is required to assess the reliability of these results.

The proposed radiomics/dosiomics-based model, using local binary pattern and topology, can be used to predict treatment failure after CRT prior to treatment planning. We believe that the proposed approach will contribute to the prediction of treatment failure in patients with PSCC.

Acknowledgements

We would like to thank Editage (www.editage.com) for English language editing.

Funding

This work was supported by JSPS KAKENHI Grant Number JP22K15808.

Conflict of interest

The authors declare that they have no conflict of interest.

Author Contributions

All the authors contributed to the conception and design of the study. The material preparation, data collection, and analysis were performed by Hidemi Kamezawa. The first draft of the manuscript was written by Hidemi Kamezawa, and all the authors commented on the previous versions of the manuscript. All authors have read and approved the final manuscript.

Ethical approval

This article does not include any studies with human participants or animals performed by any of the authors.

Kono M, Ishihara R, Kato Y et al. Diagnosis of pharyngeal cancer on endoscopic video images by Mask region-based convolutional neural network. Dig Endosc 2021;33(4):569-576. https://doi.org/10.1111/den.13800
Petit C, Lacas B, Pignon JP et al. Chemotherapy and radiotherapy in locally advanced head and neck cancer: an individual patient data network meta-analysis. Lancet Oncol 2021;22(5):727-736. https://doi.org/10.1016/S1470-2045(21)00076-0
Forastiere AA, Zhang Q, Weber RS et al. Long-term results of RTOG 91-11: a comparison of three nonsurgical treatment strategies to preserve the larynx in patients with locally advanced larynx cancer. J Clin Oncol 2013;31(7):845-852. https://doi.org/10.1200/JCO.2012.43.6097
Cooper JS, Zhang Q, Pajak TF et al. Long-term follow-up of the RTOG 9501/intergroup phase III trial: postoperative concurrent radiation therapy and chemotherapy in high-risk squamous cell carcinoma of the head and neck. Int J Radiat Oncol Biol Phys 2012;84(5):1198-1205. https://doi.org/10.1016/j.ijrobp.2012.05.008
Bernier J, Domenge C, Ozsahin M et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. N Engl J Med 2004;350(19):1945-1952. https://doi.org/10.1056/NEJMoa032641
Brockstein B, Haraf DJ, Rademaker AW et al. Patterns of failure, prognostic factors and survival in locoregionally advanced head and neck cancer treated with concomitant chemoradiotherapy: a 9-year, 337- patient, multi-institutional experience. Ann Oncol 2004;15(8):1179-1186. https://doi.org/10.1093/annonc/mdh308
Denis F, Garaud P, Bardet E et al. Final results of the 94-01 French Head and Neck Oncology and Radiotherapy Group randomized trial comparing radiotherapy alone with concomitant radiochemotherapy in advanced-stage oropharynx carcinoma. J Clin Oncol 2004;22(1):69-76. https://doi.org/10.1200/JCO.2004.08.021
Keek S, Sanduleanu S, Wesseling F et al. Computed tomography-derived radiomic signature of head and neck squamous cell carcinoma (peri)tumoral tissue for the prediction of locoregional recurrence and distant metastasis after concurrent chemo-radiotherapy. PLoS One 2020;15(5):e0232639. https://doi.org/10.1371/journal.pone.0232639, Erratum in: PLoS One. 2020;15(7):e0237048
Wu A, Li Y, Qi M et al. Dosiomics improves prediction of locoregional recurrence for intensity modulated radiotherapy treated head and neck cancer cases. Oral Oncol 2020;104:104625. https://doi.org/10.1016/j.oraloncology.2020.104625
Kamezawa H, Arimura H. Recurrence prediction with local binary pattern-based dosiomics in patients with head and neck squamous cell carcinoma. Phys Eng Sci Med 2023;46(1):99-107. https://doi.org/10.1007/s13246-022-01201-8.
Kamezawa H, Arimura H, Yasumatsu R et al. Preoperative and non-invasive approach for radiomic biomarker-based prediction of malignancy grades in patients with parotid gland cancer in magnetic resonance images. Med Imag Info Sci 2020;37(4):66-74. https://doi.org/10.11318/mii.37.66
Ikushima K, Arimura H, Yasumatsu R et al. Topology-based radiomic features for prediction of parotid gland cancer malignancy grade in magnetic resonance images. MAGMA 2023. https://doi.org/10.1007/s10334-023-01084-0. Epub ahead of print.
Le QC, Arimura H, Ninomiya K et al. Can Persistent Homology Features Capture More Intrinsic Information about Tumors from 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography Images of Head and Neck Cancer Patients? Metabolites 2022;12(10):972. https://doi.org/10.3390/metabo12100972
Nakane K, Tsuchihashi Y, Matsuura N. A simple mathematical model utilizing topological invariants for automatic detection of tumor areas in digital tissue images. Diagn Pathol 2013;8(1):1-4. https://doi.org/10.1186/1746-1596-8-S1-S27
Nakane K, Takiyama A, Mori S, Matsuura N. Homology-based method for detecting regions of interest in colonic digital images. Diagn Pathol 2015;10:1-5. https://doi.org/10.1186/s13000-015-0244-x
Pratyush P, Rien VDW, Gert V, et al. Topology and geometry of Gaussian random fields I: on Betti numbers, Euler characteristic, and Minkowski functionals. Monthly Notices of the Royal Astronomical Society, 2019;485(3):4167–4208. https://doi.org/10.1093/mnras/stz541
Grossberg A, Elhalawani H, Mohamed A et al. M.D. Anderson Cancer Center Head and Neck Quantitative Imaging Working Group. (2020) HNSCC [Dataset]. The Cancer Imaging Archive.
Grossberg A, Mohamed A, Elhalawani H et al. Imaging and Clinical Data Archive for Head and Neck Squamous Cell Carcinoma Patients Treated with Radiotherapy. Sci Data 2018;5:180173. https://doi.org/10.1038/sdata.2018.173
Elhalawani H, Mohamed A, White A et al. Matched computed tomography segmentation and demographic data for oropharyngeal cancer radiomics challenges. Sci Data 2017;4:170077. https://doi.org/10.1038/sdata.2017.77
Vallières M, Kay-Rivest E, Perrin LJ et al. (2017). Data from Head-Neck-PET-CT. The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2017.8oje5q00
Vallières M, Kay-Rivest E, Perrin LJ et al. Radiomics strategies for risk assessment of tumour failure in head-and-neck cancer. Sci Rep 2017;7(1):10117. https://doi.org/10.1038/s41598-017-10371-5.
Clark K, Vendt B, Smith K et al. The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository. J Digit Imaging 2013;26(6):1045–1057. https://doi.org/10.1007/s10278-013-9622-7
Chawla NV, Bowyer KW, Hall LO et al. SMOTE: synthetic minority over-sampling technique. Journal of artificial intelligence research, 2002;16:321-357. https://doi.org/10.1613/jair.953
Gillison ML, Trotti AM, Harris J et al. Radiotherapy plus cetuximab or cisplatin in human papillomavirus-positive oropharyngeal cancer (NRG Oncology RTOG 1016): a randomised, multicentre, non-inferiority trial. Lancet 2019;393(10166):40-50. https://doi.org/10.1016/S0140-6736(18)32779-X
Haralick RM, Shanmugam K, Dinstein I. Textural Features for Image Classification. IEEE Trans Syst Man Cybern 1973;SMC-3(6):610-621. https://doi.org/10.1109/TSMC.1973.4309314
Galloway MM. Texture classification using gray level run length. Comput Graph Image Process 1975;4:172-179. https://doi.org/10.1016/S0146-664X(75)80008-6
Thibault G, Fertil B, Navarro C et al. Texture indexes and gray level size zone matrix. application to cell nuclei classification. Proc of Pattern Recognition and Information Processing 2009:140-145.
Amadasun M, King R. Textural features corresponding to textural properties. IEEE Trans Syst Man Cybern 1989;19:1264-1274. https://doi.org/10.1109/21.44046
Mallat S. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1989;11(7):674-693. https://doi.org/10.1109/34.192463.
Ojala T, Pietikainen M, MaenpaaT. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence 2002;24(7): 971-987. https://doi.org/10.1109/TPAMI.2002.1017623
Arimura H, Soufi M, Ninomiya K, et al. Potentials of radiomics for cancer diagnosis and treatment in comparison with computer-aided diagnosis. Radiol Phys Technol 2018;11(4): 365-374.
Pavic M, Bogowicz M, Würms X et al. Influence of inter-observer delineation variability on radiomics stability in different tumor sites. Acta Oncol 2018;57(8):1070-1074. https://doi.org/10.1080/0284186X.2018.1445283.
Bufacchi A, Caspiani O, Rambaldi G et al. Clinical implication in the use of the AAA algorithm versus the AXB in nasopharyngeal carcinomas by comparison of TCP and NTCP values. Radiat Oncol 2020;15(1):150. https://doi.org/10.1186/s13014-020-01591-7

Download PDF

Version 1

posted

You are reading this latest preprint version

Radiomics- and dosiomics-based prediction of treatment failure after chemoradiotherapy for pharyngeal squamous cell carcinoma patients

Status:

Version 1

Abstract

Figures

1 Introduction

2 Methods

2.1 Overview

2.2 Patients

2.3 Feature extraction

2.3.1 Radiomics (R) feature

2.3.2 Local binary pattern (L) feature

2.3.3 Topological (T) feature

2.4 Prediction model construction and performance evaluation

3 Results

4 Discussion

5 Conclusion

Declarations

References

Status:

Version 1