Related Experiment Video
Updated: May 20, 2026

Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Assessment of bone scans in advanced prostate carcinoma using fully automated and semi-automated bone scan index
Yoshiko Takahashi1, Mana Yoshimura, Kunihito Suzuki
1Department of Radiology, Tokyo Medical University Hachioji Medical Center, Tokyo, Japan.
Objective:
As metastasis of prostate carcinoma occurs in approximately 80 % of terminal prostate carcinoma patients, the prognostic value of the prediction of prostate carcinoma by bone scintigraphy is important. We compared the automated and semi-automated bone scan index (BSI) system with extent of disease (EOD) grade if there is a possibility to substitute for EOD grading.
Materials And Methods:
We evaluated the bone scintigraphic images of 158 prostate carcinoma patients (mean age, 69.2 years old; range 50-97). Bone scans were obtained approximately 3 h after the intravenous injection of 740 MBq technetium-99 m-methylene diphosphonate. EOD grade was evaluated by 2 experienced radiologists using bone scintigraphy, magnetic resonance imaging, and computed tomography. We calculated the BSI using the Bonenavi(®) system (Fujifilm RI Pharma Co., Ltd.), utilizing data from a Japanese database. The semi-automated BSI of the patients was obtained by modifying the automated BSI independently by 3 radiologists (referred to as "observers" in this study) with 25, 10, and 4 years of experience. We then compared the EOD with the corresponding 4 independent BSIs for each patient. We used the Steel-Dwass test for multiple comparisons of the BSI among different EOD groups of patients. We analyzed the receiver-operating characteristics (ROC) curve to determine the cutoff values of sensitivity and specificity, which were both set at 95 %.
Results:
There were significant correlations observed among the mean EOD and BSI scores as determined using the Bonenavi(®) system for every patient group for all observers and the automated method. There was also a statistically significant difference in the mean BSI among all EOD groups (grades 0, 1, or 2-4) for all observers and the automated method. Each ROC curve showed an ideal shape and was within the optimal cutoff range.
Conclusion:
On the basis of the present results, BSI as calculated using the Bonenavi(®) system significantly correlated with EOD. Sensitivity and specificity as measured by the fully automated method were lower than those of semi-automated BSI with modification by radiologists. Therefore, semi-automated BSI is considered to have the possibility to substitute for EOD grading to predict the survival of prostate carcinoma patients with bone metastases, with only slight interobserver variation.
More Related Videos
06:31Quantitative [18F]-Naf-PET-MRI Analysis for the Evaluation of Dynamic Bone Turnover in a Patient with Facetogenic Low Back Pain
Published on: August 8, 2019
12:23Multi-modal Imaging of Angiogenesis in a Nude Rat Model of Breast Cancer Bone Metastasis Using Magnetic Resonance Imaging, Volumetric Computed Tomography and Ultrasound
Published on: August 14, 2012