The application of the nomogram model based on ADC histogram features in predicting clinically significant prostate cancer

doi:10.21203/rs.3.rs-4697571/v1

Download PDF

Research Article

The application of the nomogram model based on ADC histogram features in predicting clinically significant prostate cancer

https://doi.org/10.21203/rs.3.rs-4697571/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objectives

This study aimed to develop a nomogram model using ADC histogram features to predict clinically significant prostate cancer (CSPCa).

Methods

A retrospective analysis was conducted on 283 patients with suspected prostate cancer admitted to the Urology Department of Jiangnan University Affiliated Central Hospital from January 2019 to June 2024. Patients were randomly divided into a training set (70%, 198 cases) and an internal validation set (30%, 85 cases). Key features were selected through univariate analysis and LASSO regression, and a predictive model was further constructed using univariate and multivariate Logistic regression analysis. The validity of the model was assessed through ROC curves, calibration curves, and decision curve analysis.

Results

The study found that ADC_CoeffOfVar (odds ratio OR = 1.01, P = 0.034) and ADC_entropy (OR = 1.00, P < 0.001) are independent predictors for CSPCa. The nomogram model constructed based on these factors showed good predictive performance in both the training set (AUC = 0.844) and the internal validation set (AUC = 0.765). Calibration curve analysis showed that the model's predictions were highly consistent with actual observations, and decision curve analysis (DCA) further confirmed the net clinical benefit of the model in clinical decision-making.

Conclusion

The nomogram model constructed based on ADC histogram features not only provides a non-invasive tool for preoperative risk assessment but also has potential for practical clinical application.

Clinically significant prostate cancer

Magnetic resonance imaging

Histogram

Nomogram

Prostate cancer (Prostate Cancer, PCa), as the most common malignant tumor in the male urogenital system, poses a significant threat to the health of men worldwide. Statistical data show that PCa ranks second in male cancer mortality and fifth in all causes of death, with its associated cancer mortality rate exceeding 10% [1]. These data highlight the severity of prostate cancer and the urgent need for effective diagnosis and treatment strategies. Early diagnosis of prostate cancer is crucial for improving patient survival rates and quality of life [2]. With the development of medical technology, especially the application of multi-parametric magnetic resonance imaging (Multi-parametric Magnetic Resonance Imaging, mpMRI), a new perspective has been provided for the early screening and accurate diagnosis of prostate cancer [3]. mpMRI, by combining different imaging sequences, can provide more comprehensive tumor information, thereby significantly improving diagnostic accuracy [4].

In mpMRI, Diffusion Weighted Imaging (DWI) and its derived Apparent Diffusion Coefficient (ADC) are key functional imaging sequences. DWI can reveal the presence and characteristics of prostate cancer by reflecting changes in tissue microstructure [5, 6]. Changes in ADC values provide important information for assessing the biological characteristics of tumors, helping to distinguish between clinically significant and non-significant prostate cancer [7]. Clinically significant prostate cancer (Clinically Significant Prostate Cancer, CSPCa) is defined as PCa with a Gleason score ≥ 3 + 4, with or without a lesion volume ≥ 0.5cm3, with or without extracapsular invasion [8]. In contrast, non-clinically significant prostate cancer (non-CSPCa), including benign prostatic hyperplasia and clinically insignificant prostate cancer, has different treatment and management strategies from CSPCa [9]. Therefore, accurately distinguishing CSPCa from non-CSPCa is crucial for formulating personalized treatment plans.

This study aimed to explore the application of a nomogram model based on ADC histogram features in predicting CSPCa, in hopes of providing a non-invasive, efficient risk assessment tool for clinical use.

Study Subjects

A total of 283 patients with prostate cancer admitted to our hospital's urology department from January 2019 to June 2024 were collected, aged 35 to 93, with an average age of (69.89 ± 7.92) years. All patients were confirmed by surgical pathology or prostate biopsy, with 140 cases in the CSPCa group and 143 cases in the non-CSPCa group.

Inclusion criteria: MP-MRI examination was performed within one month before surgery, with clear images and no significant artifacts affecting the diagnosis; no radiotherapy, endocrine therapy, or other anti-tumor treatments were received before surgery; Exclusion criteria: Patients with severe cardiovascular and cerebrovascular diseases and severe liver and kidney function damage. All patients signed an informed consent form before entering the study.

MRI Scanning Protocol

Patients were in the supine position, with feet first, breathing calmly, and the scanning range was from the lower edge of the pubic symphysis to the iliac crest level. A laxative was taken before the examination to clean the intestines, and an appropriate amount of water was drunk to keep the bladder full. A German Siemens Magnetom Skyra 3.0 T intelligent magnetic resonance imaging device was used, with a 32-channel body phased array coil as the receiving coil. The scanning sequences included prostate T1WI (axial), T2WI (axial fat-saturated image, coronal, and sagittal), and DWI, dynamic contrast enhancement (DCE), with specific scanning parameters as shown in Table 1.

Table 1

Individual parameter values for the MRI scan sequence
scanning sequence	T1WI axis position	T2WI axis position	The T 2 WI sagittal position	The T 2 WI coronal position	DWI(b = 0s/mm 2)	DWI(b = 1000s/mm 2)	DCE
type	TSE	TSE-FS	TSE	TSE	RESOLVE	RESOLVE	t1_vibe_tra_dyn
TR/TE(ms)	744/13	4400/96	3800/117	3550/117	6570/88	6570/88	5.08/1.77
FOV(mm)	100	100	100	100	69.79	69.79	100
matrix	512×512	640×640	320×320	640×640	134×192	134×192	192×192
Floor thickness / floor spacing (mm)	3.5/4.55	3.5/4.2	3/3.6	3/3.6	3.5/4.55	3.5/4.55	3/0
NSA (times)	1	2	2	2	1	2	1
Flip Angle (degree)	160	160	160	160	180	180	15

Image Analysis

After copying all the original data, it was de-identified through a specialized de-identification software, and the de-identified data was imported into the Firevoxel software (v. 422C, https://www.firevoxel.org/) for analysis. For each patient, two physicians with more than 5 and 10 years of experience in prostate MRI diagnosis, respectively, analyzed and evaluated the data in a double-blind manner, referring to T1WI, T2WI, and DCE sequences to delineate the entire region of interest (Regions of Interest, ROI) along the edge of the tumor on the DWI (b = 1000s/mm2) image. The delineation was performed layer by layer, without delineating the top and bottom layers of the lesion to prevent partial volume effects from affecting the results. During the delineation process, areas of hemorrhage, necrosis, cystic degeneration, or calcification were avoided. After the ROI was drawn on each layer, the "Play 4D experiment" button in the software toolbar was used to synthesize the VOI of the lesion and synchronize it with the corresponding ADC image. The "ROI Stats 3D" function was used to automatically extract ADC histogram parameters. The two physicians re-measured once a month after the first measurement (see Figs. 1–2). Histogram parameters included: ADC minimum value (ADC_min), ADC maximum value (ADC_max), ADC mean (ADC_mean), ADC standard deviation (ADC_StDev), ADC variance (ADC_Variance), ADC coefficient of variation (ADC_CoeffOfVar), skewness (ADC_skewness), kurtosis (ADC_kurtosis), entropy (ADC_entropy), and the 1st, 5th, 10th, 25th, 50th, 75th, 90th, 95th, and 99th percentiles of ADC (ADC 1%, ADC 5%, ADC 10%, ADC 25%, ADC 50%, ADC 75%, ADC 90%, ADC 95%, ADC 99%).

Statistical Methods

Statistical analysis was performed using R software (version 4.1.0). Variables not normally distributed were represented by median and interquartile range and compared between groups using the Mann-Whitney U test. Categorical variables were expressed as percentages and analyzed using the chi-square test (χ² test). First, feature selection was performed using LASSO regression to reduce multicollinearity and select important predictive variables. Next, univariate and multivariate Logistic regression analysis was used to construct a nomogram model for predicting CSPCa. To evaluate the predictive performance of the model, the receiver operating characteristic (ROC) curve and its area under the curve (AUC) were used. The effectiveness of the model was verified by calculating the AUC value, as well as the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the model. The calibration curve (CC) was drawn to assess the degree of agreement between the model's predicted results and actual results. In addition, the decision curve analysis (DCA) was used to evaluate the clinical net benefit of the model. To ensure the consistency of data measurement, the intraclass correlation coefficient (ICC) was used to analyze the consistency of inter- and intra-observer measurement data. An ICC less than 0.4 indicates poor consistency, 0.4 to 0.75 indicates moderate consistency, and greater than 0.75 indicates high consistency. All statistical tests were two-sided, and a P value less than 0.05 was considered statistically significant.

Baseline Characteristics

A total of 283 patients were included in this study, divided into a training set (198 cases) and an internal validation set (85 cases) according to a 7:3 ratio. There were no statistically significant differences in age, PSA, Gleason score, and Gleason grade group (GG) between the two groups of patients (P > 0.05). See Table 2.

Table 2

Demographic characteristics of the training set and the internal validation set
Baseline metrics	training set (n = 198)	Internal validation set (n = 85)	P price
Age (interquartile spacing)	72 (66, 75)	71 (64, 75)	0.600
PSA (interquartile spacing, ng/ml)	13 (7, 30)	17 (8, 44)	0.393
GS [Example (%)]			0.562
0(nPCa)	93 (47%)	36 (42%)
6	11 (5.6%)	3 (3.5%)
7	35 (18%)	15 (18%)
8	19 (9.6%)	7 (8.2%)
9	26 (13%)	19 (22%)
10	14 (7.1%)	5 (5.9%)
GG [Example (%)]			0.673
0(nPCa)	93 (47%)	36 (42%)
1	11 (5.6%)	3 (3.5%)
2	20 (10%)	6 (7.1%)
3	15 (7.6%)	8 (9.4%)
4	19 (9.6%)	8 (9.4%)
5	40 (20%)	24 (28%)

Inter- and Intra-Observer Consistency Analysis

As shown in Table 3, the inter-observer (physician 1 vs. physician 2) and intra-observer (first measurement vs. second measurement) ICCs for the ADC histogram of csPCa were all greater than 0.9, indicating high consistency.

Table 3

Analysis of observer agreement within and between groups in the ADC histograms of prostate cancer
	Interobserver (Physician 1 vs Physician 2)		Within the observer
	The first measurement	The second measurement	Physician 1	Doctor 2
ICC	0.924	0.917	0.938	0.945
95% CI	0.889–0.956	0.886–0.963	0.901–0.966	0.911–0.981

Univariate Analysis and Feature Selection

In the univariate analysis, ADC_min, ADC_mean, ADC_StDev, ADC_Variance, ADC_CoeffOfVar, ADC_skewness, ADC_kurtosis, ADC_entropy (nats), and the ADC 1%, 5%, 10%, 25%, 50%, 75%, 90% percentiles showed statistically significant differences in the training set (P < 0.05). Through the least absolute shrinkage and selection operator (LASSO) and 10-fold cross-validation, ADC_min, ADC_CoeffOfVar, ADC_kurtosis, and ADC_entropy were selected. See Fig. 3.

Univariate and Multivariate Logistic Regression Analysis

Univariate Logistic regression analysis showed that the OR values for ADC_min, ADC_CoeffOfVar, ADC_kurtosis, and ADC_entropy were the same, with P values all less than 0.001, and AUC values of 0.764, 0.744, 0.778, and 0.816, respectively. Multivariate Logistic regression analysis showed that ADC_CoeffOfVar (OR = 1.01, P = 0.034) and ADC_entropy (OR = 1.00, P < 0.001) are independent risk factors for predicting CSPCa. The nomogram model constructed with these factors had an AUC of 0.844 in the training set and 0.765 in the internal validation set. See Tables 4–5 and Fig. 4.

Table 4

Univariate and multivariate analysis
Observational indicators	Univariate analysis			multiplicity
Observational indicators	OR	95% CI	P price	OR	95% CI	P price
ADC_min	1.00	1.00, 1.00	< 0.001	1.00	1.00, 1.00	0.889
ADC_CoefOfVar	1.01	1.01, 1.01	< 0.001	1.01	1.00, 1.01	0.034
ADC_Kurtosis	1.00	1.00, 1.00	< 0.001	1.00	1.00, 1.00	0.087
ADC_Entropy	1.00	1.00, 1.00	< 0.001	1.00	1.00, 1.00	< 0.001

Table 5

Diagnostic efficacy analysis of the nomogram model for predicting CSPCa in the training set and the internal validation set
	training set	Internal validation set
AUC	0.844	0.765
sensitivity (%)	73.4	60
specificity (%)	81.7	69.2
The positive predictive value of (%)	78.4	69.2
Negative predictive value of (%)	77.3	60
accuracy (%)	77.8	64.3

Internal Validation of Calibration Curves and Decision Curves

The calibration curves for both the training and validation sets showed that the nomogram constructed with ADC histogram features was highly consistent with the actual results, and the Hosmer-Lemeshow test showed that both the training set (P = 0.02246) and the validation set (P = 0.1202) had a high degree of fit. The DCA results indicated that compared with not using a predictive model (all patients treated or untreated), the nomogram models for the training and validation sets had additional clinical benefits. See Fig. 5.

The results of this study show that physicians 1 and 2 have high inter- and intra-observer consistency, which is consistent with previous research results [10, 11]. This finding confirms that both physicians strictly followed the delineation process, effectively avoiding potential selection bias and errors due to software operation, measurement techniques, and diagnostic experience, providing a solid foundation for the reliability of subsequent studies. The high consistency results emphasize the repeatability of our method, which is crucial for reducing variability in medical practice and improving diagnostic accuracy. In addition, this high degree of consistency also indicates the universality of our nomogram model across physicians, providing the possibility for its widespread application in different clinical settings.

Through univariate analysis, feature selection by LASSO regression, and univariate and multivariate Logistic regression analysis, this study successfully identified ADC_CoeffOfVar and ADC_entropy as key parameters for predicting CSPCa. The nomogram model constructed based on these parameters showed excellent diagnostic performance in both the training set (AUC = 0.844) and the internal validation set (AUC = 0.765), confirming its effectiveness in differentiating CSPCa from non-CSPCa. The high diagnostic efficacy of this model makes it a powerful tool for personalized screening of CSPCa, helping to avoid unnecessary biopsies and overtreatment for patients with non-CSPCa.

In clinical practice, prostate biopsy not only may lead to complications such as bleeding but also may result in repeated biopsies due to inaccurate targeting, increasing the physical and economic burden on patients [12]. Therefore, reducing unnecessary biopsies is highly attractive to both patients and clinicians. The nomogram model developed in this study provides an intuitive and concise method for assessing the necessity of biopsy, which is particularly important for the majority of patients who lack medical background knowledge.

DWI, as a commonly used functional imaging sequence for diagnosing prostate cancer, is based on the Brownian motion and diffusion attenuation of water molecules, reflecting microstructural changes in living tissues. The ADC map, calculated from two DWI images with different b values, provides richer pathophysiological information, showing higher reliability and stability [13]. Numerous studies [14–16] have confirmed that features extracted from ADC images have high accuracy in predicting PCa. For example, a study by Shaish et al. [17] showed that the accuracy of ADC image features in predicting PCa was as high as 97.39%. Although the accuracy of predicting CSPCa in this study was 77.8%, slightly lower than the aforementioned study, considering that this study only focused on CSPCa and included some patients with clinically insignificant prostate cancer with a Gleason score of 3 + 3, this difference is explainable.

With the rapid development of computer technology and artificial intelligence, radiomics has shown great potential and value in the diagnosis, differential diagnosis, treatment, and prognosis assessment of PCa. Many studies at home and abroad have developed various models for predicting CSPCa, including radiomics models, clinical parameter models, PI-RADS v2.1 models, and their combined models, all with high AUC values (all over 0.8). For example, Lu et al. [18] established models for predicting PCa and CSPCa by combining PSA and its derived indicators with PI-RADS v2, with AUC values of 0.889 and 0.925, respectively. Li et al. [19] combined radiomics features with PI-RADS v2.1, obtaining AUC values of 0.989 for the training set and 0.931 for the validation set. Xu et al. [20] showed that the radiomics model was superior to the clinical parameter model in diagnostic efficacy (AUC = 0.92 vs. 0.73, P < 0.05). Woznicki et al. [21] found that the combined model of radiomics, PI-RADS, and clinical parameters was more effective in diagnosing CSPCa and CIPCa than the PI-RADS model alone (AUC = 0.844 vs 0.688). Zhang et al. [22] developed a combined model of radiomics features and ADC values, which also showed high AUC values in distinguishing CSPCa from CIPCa (training set, internal validation set, and external validation set AUC values were 0.95, 0.93, and 0.84, respectively). The results of this study show that the AUC values of the nomogram model constructed with features extracted from ADC images were 0.844 for the training set and 0.765 for the internal validation set, higher than the AUC value of the standalone clinical parameter model, and similar to the AUC of other combined models, indicating that the diagnostic efficacy of the nomogram model constructed with ADC histogram features is high, with good discrimination ability for CSPCa, and can be used in the decision-making of prostate biopsy. For example, a 60-year-old male patient with PSA above the normal range (> 4ng/mL) was diagnosed with suspected prostate cancer by a doctor, who recommended a prostate biopsy. Before the biopsy, a prostate MRI examination was performed, and features were extracted from the ADC histogram. Assuming case 1: the prostate ADC_CoeffOfVar is 900×10 − 3 mm2/s; ADC_entropy is 4000×10 − 3 mm2/s, the risk of diagnosing CSPCa is 70%, and the patient may choose a prostate biopsy; assuming case 2: the prostate ADC_CoeffOfVar is 400×10 − 3 mm2/s, ADC_entropy is 2000×10 − 3 mm2/s, the risk of diagnosing CSPCa is less than 5%, and monitoring treatment may be chosen. Therefore, this nomogram model can compare the image and ADC value differences between two patients with similar clinical characteristics and biopsy/monitoring outcomes, thereby stratifying the risks for two clinically indistinguishable patients.

Limitations

This study is a single-center, retrospective study with a relatively small sample size and lacks external validation. Multi-center data validation is needed in the future; the lesions were not anatomically divided (transition zone, peripheral zone), and whether there are differences in lesions from different divisions needs to be verified; ROI delineation is based on manual operation, which may introduce bias, and semi-automatic or automatic ROI delineation may be used to verify the results in the future; part of the pathology in this study comes from prostate biopsies, which may not be consistent with the final surgical pathology results.

In summary, this study successfully developed a single-center nomogram model based on ADC histogram features. The model has been internally validated and shows high accuracy and reliability in predicting clinically significant prostate cancer. This finding provides a scientific basis for the decision-making of prostate biopsy and opens up a new perspective for the clinical prediction and treatment of prostate cancer, with significant clinical application value.

Acknowledgements

Not applicable.

Author contributions

Xi Gao was involved in conceiving and designing the study, conducting research, collecting and organizing data, analyzing and interpreting data, and writing the final manuscript. Shuanglin Zhang was involved in conceiving and designing the study, analyzing and interpreting data, and writing the final manuscript.

Funding

This study received no funding support.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate

This study was conducted in accordance with the Helsinki Declaration, and approval to conduct the study was obtained from the Research Committee. Informed consent was waived as the study involved a retrospective chart review. To ensure the confidentiality of the subjects, access to the completed forms and the database was limited to the research team. The data were securely stored in an office and were not shared with any unauthorized individuals.

Consent for publication

Not applicable.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Siegel RL, Giaquinto AN, Jemal A (2024) Cancer statistics, 2024. CA Cancer J Clin 74:12-49.http://doi.org/10.3322/caac.21820
Zhang Y, Li Z, Gao C, Zhang L, Huang Y, Qu H, Shu C, Wei Y, Xu M, Cui F (2024) Radiomic nomogram based on bi-parametric magnetic resonance imaging to predict the International Society of Urological Pathology grading >/= 3 prostate cancer: a multicenter study. Clin Radiol.http://doi.org/10.1016/j.crad.2024.04.011
Mohler JL, Antonarakis ES, Armstrong AJ, D'Amico AV, Davis BJ, Dorff T, Eastham JA, Enke CA, Farrington TA, Higano CS, Horwitz EM, Hurwitz M, Ippolito JE, Kane CJ, Kuettel MR, Lang JM, McKenney J, Netto G, Penson DF, Plimack ER, Pow-Sang JM, Pugh TJ, Richey S, Roach M, Rosenfeld S, Schaeffer E, Shabsigh A, Small EJ, Spratt DE, Srinivas S, Tward J, Shead DA, Freedman-Cass DA (2019) Prostate Cancer, Version 2.2019, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw 17:479-505.http://doi.org/10.6004/jnccn.2019.0023
Chen Y, Xu D, Ruan M, Li H, Lin G, Song G (2023) A prospective study of the prostate health index density and multiparametric magnetic resonance imaging in diagnosing clinically significant prostate cancer. Investig Clin Urol 64:363-372.http://doi.org/10.4111/icu.20230060
Hamm CA, Baumgartner GL, Biessmann F, Beetz NL, Hartenstein A, Savic LJ, Frobose K, Drager F, Schallenberg S, Rudolph M, Baur A, Hamm B, Haas M, Hofbauer S, Cash H, Penzkofer T (2023) Interactive Explainable Deep Learning Model Informs Prostate Cancer Diagnosis at MRI. Radiology 307:e222276.http://doi.org/10.1148/radiol.222276
Lu JY, Yu H, Zou XL, Li Z, Hu XM, Shen YQ, Hu DY (2019) Apparent diffusion coefficient-based histogram analysis differentiates histological subtypes of periampullary adenocarcinoma. World J Gastroenterol 25:6116-6128.http://doi.org/10.3748/wjg.v25.i40.6116
Sun Z, Wu P, Cui Y, Liu X, Wang K, Gao G, Wang H, Zhang X, Wang X (2023) Deep-Learning Models for Detection and Localization of Visible Clinically Significant Prostate Cancer on Multi-Parametric MRI. J Magn Reson Imaging 58:1067-1081.http://doi.org/10.1002/jmri.28608
Ueda T, Ohno Y, Shinohara M, Yamamoto K, Ikedo M, Yui M, Yoshikawa T, Takenaka D, Ishida S, Furuta M, Matsuyama T, Nagata H, Ikeda H, Ozawa Y, Toyama H (2023) Reverse encoding distortion correction for diffusion-weighted MRI: Efficacy for improving image quality and ADC evaluation for differentiating malignant from benign areas in suspected prostatic cancer patients. Eur J Radiol 162:110764.http://doi.org/10.1016/j.ejrad.2023.110764
Lee SI, Hectors SJ (2020) Prostate MRI: Toward Imaging Tumor Histology. Radiology 296:356-357.http://doi.org/10.1148/radiol.2020202064
Zhang DD, You C, Gu YJ (2021) Observer consistency study based on dynamic contrast-enhanced MRI radiomics parameters. Tumor Imaging 30:252-256. http://doi.org/10.19732/j.cnki.2096-6210.2021.04.004
You C, Li J, Zhi W, Chen Y, Yang W, Gu Y, Peng W (2019) The volumetric-tumour histogram-based analysis of intravoxel incoherent motion and non-Gaussian diffusion MRI: association with prognostic factors in HER2-positive breast cancer. J Transl Med 17:182.http://doi.org/10.1186/s12967-019-1911-6
Edwin M Chau BRAS, Mieke Van Hemelrijck SMTP, Sidath H Liyanage JAVJ, Acher AP (2023) MRI-based nomogram for the prediction of prostate cancer diagnosis: A multi-centre validated patient – physician decision tool. Journal of Clinical Urology 16:588-595.http://doi.org/10.1177/20514158211065949
Xing P, Chen L, Yang Q, Song T, Ma C, Grimm R, Fu C, Wang T, Peng W, Lu J (2021) Differentiating prostate cancer from benign prostatic hyperplasia using whole-lesion histogram and texture analysis of diffusion- and T2-weighted imaging. Cancer Imaging 21:54.http://doi.org/10.1186/s40644-021-00423-5
Caglic I, Sushentsev N, Syer T, Lee KL, Barrett T (2024) Biparametric MRI in prostate cancer during active surveillance: is it safe? Eur Radiol.http://doi.org/10.1007/s00330-024-10770-z
Spinner JW, Purysko AS, Westphalen AC (2024) Enhancing prostate MRI expertise: educational strategies for radiologists. Abdom Radiol (NY).http://doi.org/10.1007/s00261-024-04325-5
Ogreden E, Oguz U, Demirelli E, Tok DS, Aslan S, Tok B, Ciftci OC, Akyol S, Altun E, Ahmadli C, Semiz U, Batur H (2024) The role of multiparametric magnetic resonance imaging in the diagnosis of granulomatous prostatitis mimicking prostate cancer. Abdom Radiol (NY).http://doi.org/10.1007/s00261-024-04288-7
Shaish H, Kang SK, Rosenkrantz AB (2017) The utility of quantitative ADC values for differentiating high-risk from low-risk prostate cancer: a systematic review and meta-analysis. Abdom Radiol (NY) 42:260-270.http://doi.org/10.1007/s00261-016-0848-y
Lu YF, Zhang Q, Yao WG, Chen HY, Chen JY, Xu CC, Yu RS (2019) Optimizing prostate cancer accumulating model: combined PI-RADS v2 with prostate specific antigen and its derivative data. Cancer Imaging 19:26.http://doi.org/10.1186/s40644-019-0208-6
Li M, Yang L, Yue Y, Xu J, Huang C, Song B (2020) Use of Radiomics to Improve Diagnostic Performance of PI-RADS v2.1 in Prostate Cancer. Front Oncol 10:631831.http://doi.org/10.3389/fonc.2020.631831
Xu M, Fang M, Zou J, Yang S, Yu D, Zhong L, Hu C, Zang Y, Dong D, Tian J, Fang X (2019) Using biparametric MRI radiomics signature to differentiate between benign and malignant prostate lesions. Eur J Radiol 114:38-44.http://doi.org/10.1016/j.ejrad.2019.02.032
Woznicki P, Westhoff N, Huber T, Riffel P, Froelich MF, Gresser E, von Hardenberg J, Muhlberg A, Michel MS, Schoenberg SO, Norenberg D (2020) Multiparametric MRI for Prostate Cancer Characterization: Combined Use of Radiomics Model with PI-RADS and Clinical Parameters. Cancers (Basel) 12.http://doi.org/10.3390/cancers12071767
Zhang Y, Chen W, Yue X, Shen J, Gao C, Pang P, Cui F, Xu M (2020) Development of a Novel, Multi-Parametric, MRI-Based Radiomic Nomogram for Differentiating Between Clinically Significant and Insignificant Prostate Cancer. Front Oncol 10:888.http://doi.org/10.3389/fonc.2020.00888

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

The application of the nomogram model based on ADC histogram features in predicting clinically significant prostate cancer

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods

Study Subjects

MRI Scanning Protocol

Image Analysis

Statistical Methods

Results

Baseline Characteristics

Inter- and Intra-Observer Consistency Analysis

Univariate Analysis and Feature Selection

Univariate and Multivariate Logistic Regression Analysis

Internal Validation of Calibration Curves and Decision Curves

Discussion

Limitations

Conclusions

Declarations

References

Additional Declarations

Status:

Version 1