PET/CT Radiomics Integrated with Clinical Indexes as A Tool to Predict Ki67 in Breast Cancer: A Pilot Study

doi:10.21203/rs.3.rs-3145802/v1

Download PDF

Research Article

PET/CT Radiomics Integrated with Clinical Indexes as A Tool to Predict Ki67 in Breast Cancer: A Pilot Study

https://doi.org/10.21203/rs.3.rs-3145802/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background:

Breast cancer (BC) represents the second cause of cancer-related death among women globally, and Ki67 was demonstrated as an important predictive biomarker in worse survival and neoadjuvant treatment in BC. This study aims to investigate the value of radiomics features derived from ¹⁸F-FDG PET/CT combined with clinical characteristics in predicting Ki67 in patients with BC.

Methods:

A total of 114 patients diagnosed as BC and examined using ¹⁸F-FDG PET/CT were included in this study. Patients were randomly separated into a training set (n = 79, with 55 cases of Ki67 + and 24 cases of Ki67-) and a validation set (n = 35, with 24 cases of Ki67 + and 11 cases of Ki67-) at a ratio of 7:3. Thirteen clinical characteristics and 704 radiomics features were extracted, and the univariance logistic analysis, max-Relevance and Min-Redundancy, the least absolute shrinkage and selection operator regression, and the Spearman test were applied for feature selection. Three models were developed, including the clinical model, the radiomics model, and the combined model, and a nomogram of the combined model was constructed. The predictive performance of all three models was examined by the receiver operating characteristic (ROC) curve. Clinical utility was validated by decision curve analysis (DCA).

Results:

The N stage, tumor morphology, maximal standard uptake value, and the longest diameter were significantly different in Ki67 + and Ki67- groups (P < 0.05) and were selected as the most discriminative clinical features. Eight radiomics features were selected for the radiomics model. In total, 7 radiomics and the above 4 clinical characteristics were selected for the combined model. The AUC of the combined model in the training and test group was 0.90 (95% Confidence Interval (CI): 0.82–0.97) and 0.81 (95% CI: 0.64–0.99), respectively. The combined model significantly outperformed the radiomics model and the clinical model alone (P < 0.05). The DCA curve showed the advantages of the combined model over the clinical model and radiomics model.

Conclusions:

The radiomics-derived features combined with the clinical features could effectively predict Ki67 expression in BC based on PET/CT images.

Trial registration:

This study was registered at ClinicalTrials Gov (number NCT05826197) on 7th, May 2023.

PET/CT

radiomics

breast cancer

Ki67 prediction

model

Breast cancer (BC) is a heterogeneous malignancy for women, and BC represents the second cause of cancer-related death among women globally, with increasing incidence over the last decade¹. Ki67 is the most commonly used biomarker for evaluating the proliferative index of BC. Several studies have found that high Ki67 expression is associated with an elevated rate of relapse and worse survival in BC ^2–5. Other studies have demonstrated the clinical validity of Ki67 as a predictive marker in neoadjuvant settings⁶. Nevertheless, tumor biology characterization is an invasive procedure, thus a non-invasive technique for Ki67 evaluation would be worthwhile.

¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography/computed tomography (PET/CT) is widely and routinely used in BC staging⁷. ¹⁸F-FDG PET/CT can provide extra metabolic information compared with traditional imaging modalities. Standard semi-quantitative imaging variables obtained from ¹⁸F-FDG PET/CT, such as the maximum standardized uptake value (SUVmax), the mean standardized uptake value (SUVmean), and the metabolic tumor volume (MTV), have been shown to correlate with tumor aggressiveness and patient outcomes in BC^8–11. But those quantitative variables were not able to fully capture BC heterogeneity, for example, SUVmax represents only the single hottest pixel.

Radiomics, an approach that aims to extract quantitative features from medical imaging to quantify lesion heterogeneity and texture, holds promise in addressing clinical challenges in monitoring disease progression¹². Recent reports indicate that features obtained from ¹⁸F-FDG PET/CT are associated with the molecular subtypes^13–15, but limited evidence relating to their roles with Ki67.

The main objective of this study was to investigate the value of radiomics features derived from ¹⁸F-FDG PET/CT combined with clinical characteristics in predicting Ki67 in patients with BC.

2.1 Study population

This retrospective study was conducted at the First Affiliated Hospital of Xi 'an Jiaotong University (NCT05826197), and the study protocol was approved by the Ethics Committee of Xi 'an Jiaotong University (IRB-SOP-AF-16).

Between November 2016 to June 2020, a total of 129 female patients with BC who underwent ¹⁸F-FDG PET/CT examinations were retrospectively studied. Inclusion criteria were: 1) BC confirmed through preoperative puncture or postoperative pathology; 2) underwent ¹⁸F-FDG PET/CT examination; and 3) available Ki67 expression. Exclusion criteria were: 1) the primary lesion was too small to detect by ¹⁸F-FDG PET/CT or occult BC patients (n = 5); 2) diffuse lesion on unilateral side or multifocal lesions in bilateral side (n = 4); 3) benign breast lesions (n = 1); and 4) neo-adjuvant chemotherapy or anti-tumor treatment performed before imaging (n = 4).

A flowchart of this process is shown in Fig. 1. A total of 114 patients were enrolled in the study. Patients were randomly assigned to the training group (79 patients) and test group (35 patients) at a ratio of 7 to 3 using a stratified sampling method to ensure the balance of positive and negative samples in both groups. The age, location (right side or left side) of the BC, and the menopausal status (premenopausal, or during and beyond menopausal) of the patient were recorded.

2.2 Immunohistochemistry

Formalin-fixed paraffin-embedded tissue samples from BC cases were used for Ki67 assessment performed by an experienced pathologist who was blinded to the PET/CT results. Ki67 levels were divided into the 0.5%, 10%, 15%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, and 100% groups according to the degree of expression, and an expression index ≥ 20% was considered positive.

2.3 PET/CT data acquisition

All examinations were performed using a 64-detector scanner (Gemini TF PET/CT, Philips, Netherlands). ¹⁸F-FDG was synthesized using a small cyclotron (GE MINItrace) and an FDG synthesis module. Radiochemical purity was > 99%. Both endotoxin and bacteriological tests were negative, which met the radiopharmaceutical requirements.

The patients fasted for more than 6 h before intravenous injection of ¹⁸F-FDG (3.7 MBq/kg). The fasting blood glucose level should be lower than 12.0 mmol/L. After resting for 60 min, the patients underwent whole-body PET/CT. The scan scope was from the top of the skull or the level of the first thoracic vertebra to the upper femur. PET collects 6–10 beds with 1.5 min/bed. CT scans (tube voltage, 120 kV; automatic milliampere second; matrix, 512×512; layer thickness, 5 mm) were performed for lesions’ location and attenuation correction of the PET image. MIP (Maximum Intensity Projection), PET, CT, and fusion images were displayed on the Extended Brilliance Workstation (EBW) workstation.

2.4 Image analysis

Manually-defined features, including tumor morphology (regular, or irregular), necrosis (with or without), and calcification (with or without), as well as the N (N0, or N1) and M (M0, or M1) stages, were determined in a double-blind manner by two experts (Hui Ding, reader 1; Cong Shen, reader 2) with more than 10 years of PET/CT interpretation experience. Any disagreement between the two radiologists was resolved by another experienced radiologist.

The longest diameter (mm) of the cancer was measured at the maximal horizontal position. The volume of interest (VOI) was automatically delineated with a 40% maximum standardized uptake value (SUVmax) as the threshold on the EBW workstation. The VOIs were reviewed and allowed manual correction when the tumor border was not satisfied. SUVmax, mean SUV (SUVmean), standard deviation (SD) of SUV, and metabolic tumor volume (MTV) were calculated, see Fig. 2.

2.5 Feature extraction

The VOIs were saved as nifty files. Radiomics feature extraction was implemented using the Philips Radiomics Tool (Philips Healthcare, China), and the core feature calculation was based on pyRadiomics (3.0.1)¹⁶. A total of 704 three-dimensional (3D) radiomics features were extracted, including original direct (n = 83), wavelet transforms (n = 552), and logarithm transforms (n = 69). Details of the feature extraction are shown in the https://pyradiomics.readthedocs.io/en/latest/features.html#. There were no missing data for clinical and radiomics features.

2.6 Feature reduction

In the training group, for all thirteen clinical features (general characteristics and manually measured features), a univariate logistic regression analysis test was applied to select features with a P value < 0.1 for the subsequent analysis. The maximum relevance minimum redundancy (mRMR, top 30 features were selected) and least absolute shrinkage and selection operator (LASSO) with the optimal 𝝺 were applied to select the most discriminative radiomics features for predicting Ki67 status. After that, the Spearman test with a threshold of 0.8 was used for the clinical features and radiomics features respectively to delete the collinear features. Then, the selected clinical features and selected radiomics features were combined, and the LASSO regression model was applied again to delete the collinear features.

2.6 Statistical analysis, model construction, and evaluation

Data were analyzed using SPSS® v. 25.0. (IBM Corp., New York, NY, USA) and python V 3.9 (URL https://www.python.org/). Continuous variables with abnormal distributions were expressed as median [25%, 75%] and were tested using the Mann–Whitney U test. Categorical variables were compared using the χ² test or Fisher’s exact test. Statistical significance was set at p < 0.05.

Three models, including the clinical model, the radiomics model, and the combined clinical-radiomics model were constructed using a logistic regression model. Models were assessed using the receiver operating characteristic (ROC) curve for discriminating Ki67 + cases from Ki67- cases. Indexes including area under the curve (AUC), sensitivity, specificity, and accuracy (ACC) were calculated. The DeLong test was used to compare the difference between ROC curves.

The calibration curves and Hosmer-Lemeshow test were utilized to assess the agreement between predicted and actual probabilities of various models, and a P value > 0.05 mean good consistency. Nomogram was constructed to visualize the predictive model for Ki67 expression. The decision curve analysis (DCA) was used to was conducted to determine the clinical usefulness of the nomogram by quantifying the net benefits at different threshold probabilities.

3.1 Clinical characteristics

The clinical characteristics of the patients are summarized in Table 1. There were no significant differences between the two cohorts in terms of age, lesion location, menstruation status, or M stage (P > 0.05). In addition, tumor morphology, N stage, the longest diameter, SUVmax, SUVmean, SD, and MTV were significantly different between the two groups (P < 0.05) (Table 1).

Table 1

Clinical characteristics and manual quantitative analysis of primary tumors on PET/CT images
Parameter	All (n = 114)	Ki67- (n = 35)	Ki67+ (n = 79)	Z/χ²	p
Age (year)	49.0 [40.0; 56.0]	51.0 [40.5;60.5]	49.0 [39.5;54.0]	0.549^a	0.521
Menstrual status				0.327^b	0.567
premenopausal	67(58.3%)	19(54.3%)	48(60.0%)
during and beyond menopausal	48(41.7%)	16(45.7%)	32(40.0%)
lesion location				0.495^b	0.482
left	61 (53.5%)	17 (48.6%)	44 (55.7%)
right	53 (46.5%)	18 (51.4%)	35 (44.3%)
Tumor morphology				9.048^b	0.003
Quasi circular	64 (56.1%)	27 (77.1%)	37 (46.8%)
Irregular	50 (43.9%)	8 (22.9%)	42 (53.2%)
Necrosis				--	0.310
No	110 (96.5%)	35 (100%)	75 (94.9%)
Yes	4(3.51%)	0 (0.00%)	4 (5.06%)
Calcification				0.213^b	0.645
No	109 (95.6%)	33 (94.3%)	76 (96.2%)
Yes	5 (4.39%)	2 (5.71%)	3 (3.80%)
N stage				15.584^b	< 0.001
N0	47 (41.2%)	24 (68.6%)	23 (29.1%)
N1	67 (58.8%)	11 (31.4%)	56 (70.9%)
M stage				--	0.102
M0	102 (89.5%)	34 (97.1%)	68 (86.1%)
M1	12 (10.5%)	1 (2.86%)	11 (13.9%)
The longest diameter (mm)	18.0 [13.1;29.0]	15.1 [11.9;18.4]	20.5 [14.9;35.5]	2.520^a	0.012
SUV_max	6.32 [4.71;9.36]	5.32 [3.60;6.47]	7.29 [5.28;11.0]	5.400^a	< 0.001
SUV_mean	3.67 [2.78;5.33]	2.95 [1.93;3.85]	4.29 [2.92;5.87]	5.400^a	< 0.001
SD	1.03 [0.55;1.59]	0.57 [0.40;1.15]	1.15 [0.69;1.83]	5.400^a	< 0.001
MTV(mm³)	5824 [2896;13216]	3072 [2112;8192]	6912 [3424;16000]	2.160^a	0.031
Note: a, Z test; b, χ²; --, Fisher exact probability test; SUV, standardized uptake value; SD, standard deviation, MTV, metabolic tumor volume.

3.2 Results of feature extraction

Table 2 showed the selected clinical features after the application of univariate analysis and Spearman test, as well as the features’ weight in the logistic regression model. Thus, the clinical model = 1.203 * N stage + 0.350 * Tumor morphology + 0.244 * SUVmax + 0.037 * The longest diameter − 2.257.

For 704 radiomics features, mRMR was used firstly to select the top 30 features, then LASSO was applied for further features reduction, see Fig. 3. Table 3 showed the 8 selected radiomics features after the application of mRMR and LASSO, and the Spearman test, as well as the features’ weight in the logistic regression model. Thus, the Radscore = (-0.215) * HHH_glszm_GrayLevelNonUniformityNormalized + (-0.472) * HHL_ngtdm_Contrast + 0.313 * LHH_glszm_SmallAreaLowGrayLevelEmphasis + (-0.418) * HLL_glszm_SmallAreaEmphasis + 0.319 * LHH_gldm_DependenceVariance + 0.432 * -LLL_firstorder_90Percentile + (-0.286) * log-sigma-6-0-mm-3D_glrlm_ShortRunHighGrayLevelEmphasis + (-0.517) * LHL_glrlm_RunEntropy + 1.036.

Table 2

retained clinical features and its weight in the clinical logistic regression model
method	univariate logistic regression	Spearman test	weight
Clinical features	N stage	N stage	1.203
	SUVmax	SUVmax	0.244
	SUVmean	The longest diameter	0.037
	The longest diameter	Tumor morphology	0.350
	Tumor morphology
intercept			-2.257

SUVmax: maximum standardized uptake value; SUVmean: mean standardized uptake value.

Table 3

retained radiomics features and their weight in the radiomics logistic regression model
method	mRMR and LASSO	Spearman test	weight
Radiomics features	LHL_glrlm_RunEntropy	LHL_glrlm_RunEntropy	-0.517
	HHL_ngtdm_Contrast	HHL_ngtdm_Contrast	-0.472
	HLL_glszm_SmallAreaEmphasis	HLL_glszm_SmallAreaEmphasis	-0.418
	HHH_glszm_GrayLevelNonUniformityNormalized	HHH_glszm_GrayLevelNonUniformityNormalized	-0.215
	LHH_glszm_SmallAreaLowGrayLevelEmphasis	LHH_glszm_SmallAreaLowGrayLevelEmphasis	0.313
	LHH_gldm_DependenceVariance	LHH_gldm_DependenceVariance	0.319
	LLL_firstorder_90Percentile	LLL_firstorder_90Percentile	0.432
	log-sigma-6-0-mm-3D_glrlm_ShortRunHighGrayLevelEmphasis	log-sigma-6-0-mm-3D_glrlm_ShortRunHighGrayLevelEmphasis	-0.286
	original_firstorder_90Percentile
Intercept			1.036
mRMR: maximum relevance minimum redundancy; LASSO: least absolute shrinkage and selection operator

3.3 Construction of a combined model for predicting Ki67

The selected clinical features and Radscore were combined to construct a logistic regression model, and the results were visualized in the nomogram in Figure. 4. The calibration curve of the combined model for Ki67 expression prediction probability shows good accordance between prediction and observation in the training and test cohort (P > 0.05), shown in Fig. 5.

3.4 Evaluation of the performance of the three models

In both training and testing groups, the combined model showed satisfactory performance in predicting Ki67 (Table 4). The AUC values of the training and test groups were 0.90 (95% CI: 0.82–0.97) and 0.81 (95% CI: 0.64–0.99), respectively, see Fig. 6. The AUC values of the clinical model, radiomics model, and radiomics model were compared using the DeLong test. It was found that the combined radiomics-clinical model significantly outperformed the radiomics model and the clinical model alone (P < 0.05).

The DCA curve was used to assess the clinical usefulness of the nomogram, see Fig. 7. Under the most threshold probability, using the combined model to predict Ki67 status added more benefit than either the treat-all scheme or the treat-none scheme. The DCA curve showed the advantages of the combined model over the clinical model and radiomics model only.

Table 4

The performance of three models in the training set and test set
models	Data set	AUC (95%CI)	ACC	Sensitivity	Specificity
Clinical model	Training set	0.83[0.74 0.93]	0.81	0.82	0.79
Clinical model	Test set	0.73[0.55 0.91]	0.74	0.79	0.64
Radiomics model	Training set	0.83[0.74 0.92]	0.76	0.75	0.79
Radiomics model	Test set	0.75[0.56 0.95]	0.71	0.88	0.36
Combined model	Training set	0.90[0.82 0.97]	0.81	0.75	0.96
Combined model	Test set	0.81[0.64 0.99]	0.80	0.83	0.73

AUC: area under the curve; CI: confidence interval; ACC: accuracy. The combined model represents the model that combined the clinical and radiomics features.

Breast cancer is a heterogeneous malignancy, with high mortality and morbidity. Ki67 is important for the identification of patients with worse survival and for neoadjuvant treatment. Radiomics is a newly introduced image-analysis method involving high-throughput feature extraction from radiographic images, which is promising for the prediction of tumor heterogeneity because of its noninvasive and repeatable properties.

Our study showed that multiple semi-quantitative parameters on ¹⁸F-FDG PET/CT images correlate with Ki67 expression, which is in accordance with the previous studies^17,18. Therefore, we constructed a nomogram incorporating seven radiomics and four clinical features for predicting Ki67 expression in BC. The validation of the nomogram showed that it had good discriminative and calibration capabilities. The AUC of the training and test groups were 0.90 (95% CI: 0.82–0.97) and 0.81 (95% CI: 0.64–0.99), respectively. The DeLong test found that the combined model significantly outperformed the radiomics model and the clinical model alone. Also, the DCA displayed a greater net benefit for the combination of clinical and radiomics features compared to either of them alone. This strategy may have clinical implications for individualized follow-up and guide therapeutic strategies.

Previous studies that extracted radiomics features from various imaging modalities showed a good AUC in the prediction of Ki67 status. Fan et al found that radiomics features extracted from contrast-enhanced CT showed an AUC of 0.726 in the prediction of Ki-67 status¹⁹. Feng, S et al extracted radiomics features from dynamic contrast-enhanced magnetic resonance imaging, and their radiomics models reached an AUC of 0.839 (95% confidence interval [CI], 0.768–0.895) within the training set and 0.795 (95% CI, 0.674–0.887) within the independent validation set in the prediction of Ki-67 status²⁰. Liu et al constructed a radiomics model based on ultrasound images, and their results showed an AUC of 0.821 (95% CI:0.764–0.880) and 0.713 (95% CI:0.612–0.814) in the training and validation cohorts²¹. For our study, the metabolic information derived from PET/CT images were expressed as multiple radiomics features, which were used to predict the Ki67 status, and showed an AUC of 0.83[0.74 0.92] within the training set and 0.75[0.56 0.95] within the testing set.

Some studies reported that the integration of clinical features into radiomics features can reach higher accuracy. Wu, J et al combined clinical features and radiomics features derived from ultrasound images, and the combined model showed a higher AUC than the radiomics in the prediction of Ki67 status ²². In our study, we also showed significantly higher accuracy in the combined model, which means the interpretation of physicians is also important besides the radiomics analysis.

This study has limitations. First, the retrospective nature and its small sample size. Second, BC is a heterogeneous disease that comprises different molecular sub-types, whether Ki67 can be combined with radiomics features in the prediction of auxiliary lymph node metastasis and molecular subtype were also widely concerned.

The radiomics-derived evaluation score combined with the clinical features could effectively predict Ki67 expression in BC based on PET/CT images.

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the First Affiliated Hospital of Xi'an Jiaotong University (IRB-SOP-AF-16). All experiments were performed by the Declaration of Helsinki. All the patients provided the written informed consent of the participate.

Consent for publication

Not Applicable.

Availability of data and materials

We confirm that all the materials and data with regard to the analysis in the manuscript are available for request from the Corresponding Author (Cong Shen, [email protected])

Competing interests

The authors declare that they have no competing interests.

Funding

This study was funded by the National Natural Science Foundation of China (No. 82272073), the Key Research and Development Plan of Shaanxi Province (2023-YBSF-480), and the Natural science and basic research project of Shaanxi Province (2023-JC-QN-0903).

Authors' contributions

Cong Shen contributed to the conception of the study; Hui Ding, Cong Shen, Yan Li collected the data and measured the manually parameters; Cong Shen, Ting Liang, Yuting Liao and Xiao Yu contributed to data analysis. All authors read and approved the final manuscript.

Acknowledgments

Not applicable.

Siegel RL, Miller KD, Jemal A, Cancer statistics. 2020. CA: a cancer journal for clinicians. 2020;70(1):7–30.
Baker E, Whiteoak N, Hall L, France J, Wilson D, Bhaskar P, Mammaglobin-A. VEGFR3, and Ki67 in Human Breast Cancer Pathology and Five Year Survival. Breast cancer: basic and clinical research. 2019;13:1178223419858957.
Gallardo A, Garcia-Valdecasas B, Murata P, et al. The inverse relationship between Ki67 and survival in early luminal breast cancer: confirmation in a multivariate analysis. Breast Cancer Res Treat. 2018;167(1):31–7.
Kilickap S, Kaya Y, Yucel B, Tuncer E, Babacan NA, Elagoz S. Higher Ki67 expression is associates with unfavorable prognostic factors and shorter survival in breast cancer. Asian Pac J cancer prevention: APJCP. 2014;15(3):1381–5.
Ishihara M, Mukai H, Nagai S, et al. Retrospective analysis of risk factors for central nervous system metastases in operable breast cancer: effects of biologic subtype and Ki67 overexpression on survival. Oncology. 2013;84(3):135–40.
Acs B, Zambo V, Vizkeleti L, et al. Ki-67 as a controversial predictive and prognostic marker in breast cancer patients treated with neoadjuvant chemotherapy. Diagn Pathol. 2017;12:1–12.
Chakraborty D, Basu S, Ulaner GA, Alavi A, Kumar R. Diagnostic Role of Fluorodeoxyglucose PET in Breast Cancer: A History to Current Application. PET Clin. 2018;13(3):355–61.
Kitajima K, Miyoshi Y, Sekine T, et al. Harmonized pretreatment quantitative volume-based FDG-PET/CT parameters for prognosis of stage I-III breast cancer: Multicenter study. Oncotarget. 2021;12(2):95–105.
Kitajima K, Miyoshi Y, Sekine T, et al. Harmonized pretreatment quantitative volume-based (18)F-FDG PET/CT parameters for stage IV breast cancer prognosis. Multicenter study in Japan. Hell J Nucl Med. 2020;23(3):272–89.
Gelezhe PB, Blokhin IA, Marapov DI, Morozov SP. Quantitative parameters of MRI and (18)F-FDG PET/CT in the prediction of breast cancer prognosis and molecular type: an original study. Am J Nucl Med Mol Imaging. 2020;10(6):279–92.
Qu YH, Long N, Ran C, Sun J. The correlation of (18)F-FDG PET/CT metabolic parameters, clinicopathological factors, and prognosis in breast cancer. Clin translational oncology: official publication Federation Span Oncol Soc Natl Cancer Inst Mexico. 2021;23(3):620–7.
Tagliafico AS, Piana M, Schenone D, Lai R, Massone AM, Houssami N. Overview of radiomics in breast cancer diagnosis and prognostication. Breast. 2020;49:74–80.
Waks AG, Winer EP. Breast Cancer Treatment A Review. Jama-Journal Of the American Medical Association. 2019;321(3):288–300.
Antunovic L, Gallivanone F, Sollini M, et al. [(18)F]FDG PET/CT features for the molecular characterization of primary breast tumors. Eur J Nucl Med Mol Imaging. 2017;44(12):1945–54.
Khare S, Singh SS, Irrinki S, et al. (18)F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography Features in Locally Advanced Breast Cancer and Their Correlation with Molecular Subtypes. Indian J nuclear medicine: IJNM : official J Soc Nuclear Med India. 2018;33(4):290–4.
van Griethuysen JJM, Fedorov A, Parmar C, et al. Computational Radiomics System to Decode the Radiographic Phenotype. Cancer Res. 2017;77(21):e104–7.
Mohamadien NRA, Sayed MHM. Correlation between semiquantitative and volumetric 18F-FDG PET/computed tomography parameters and Ki-67 expression in breast cancer. Nucl Med Commun. 2021.
Deng SM, Zhang W, Zhang B, Chen YY, Li JH, Wu YW. Correlation between the Uptake of F-18-Fluorodeoxyglucose (F-18-FDG) and the Expression of Proliferation-Associated Antigen Ki-67 in Cancer Patients: A Meta-Analysis. PLoS ONE. 2015;10(6):e0129028.
Fan Y, Pan X, Yang F, et al. Preoperative Computed Tomography Radiomics Analysis for Predicting Receptors Status and Ki-67 Levels in Breast Cancer. Am J Clin Oncol. 2022;45(12):526–33.
Feng S, Yin J. Radiomics of dynamic contrast-enhanced magnetic resonance imaging parametric maps and apparent diffusion coefficient maps to predict Ki-67 status in breast cancer. Front Oncol. 2022;12:847880.
Liu J, Wang X, Hu M, et al. Development of an ultrasound-based radiomics nomogram to preoperatively predict Ki-67 expression level in patients with breast cancer. Front Oncol. 2022;12:963925.
Wu J, Fang Q, Yao J, et al. Integration of ultrasound radiomics features and clinical factors: A nomogram model for identifying the Ki-67 status in patients with breast carcinoma. Front Oncol. 2022;12:979358.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

PET/CT Radiomics Integrated with Clinical Indexes as A Tool to Predict Ki67 in Breast Cancer: A Pilot Study

Status:

Version 1

Abstract

Background:

Methods:

Results:

Conclusions:

Trial registration:

Figures

1. Background

2. Materials and Methods

2.1 Study population

2.2 Immunohistochemistry

2.3 PET/CT data acquisition

2.4 Image analysis

2.5 Feature extraction

2.6 Feature reduction

2.6 Statistical analysis, model construction, and evaluation

3. Results

3.1 Clinical characteristics

3.2 Results of feature extraction

3.3 Construction of a combined model for predicting Ki67

3.4 Evaluation of the performance of the three models

4. Discussion

5. Conclusion

Declarations

References

Additional Declarations

Status:

Version 1