Seungeun Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Seungeun Lee has a total of 54 co-authors. Here are the top co-authors by number of shared publications:
- Joon-Yong Jung
Department of Radiology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea , Seoul
10 shared publications - So-Yeon Lee
Department of Radiology, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea
10 shared publications - Sungwon Lee
Department of Radiology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea , Seoul
4 shared publications - Hyemin Park
Department of Radiology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea , Seoul
2 shared publications - Hyeondeok Choi
Department of Radiology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 06591, Republic of Korea
2 shared publications - Hyerim Park
Department of Radiology, Soonchunhyang University College of Medicine, Cheonan Hospital, Cheonan, Korea
2 shared publications
Latest Publications
- 19 October 2026
- Published by Oxford University Press (OUP) in British Journal of Radiology
Integrating diffusion-weighted imaging parameters with RECIST 1.1 for early assessment of systemic therapy response in lung cancer pelvic bone metastases
Objectives: To determine preliminary quantitative ADC thresholds for functional response and to evaluate whether integrating DWI metrics with RECIST 1.1 provides complementary information for early systemic therapy response assessment in bone metastases. Methods: This retrospective single-center study included lung cancer patients with pelvic bone metastases who underwent 3 MRIs between November 2015 and November 2023: baseline, early time point ( 1300 µm2/s or ΔADC > 0.14. An integrated response scheme (DWI-RECIST) combined this DWI-based response with RECIST assessment. Early responses at TP1 were compared between RECIST and DWI-RECIST using RECIST at TP2 as reference standard. Interobserver agreement was assessed using intraclass correlation coefficient, and diagnostic performance was evaluated using net reclassification improvement and AUCs. Results: Seventy-eight lesions from 27 patients were included. Compared with RECIST, DWI-RECIST yielded net classification gains of 7.4% (reader 1) and 11.1% (reader 2). DWI-RECIST showed higher AUCs than RECIST (reader 1: 0.6964 vs. 0.6184, P = .472; reader 2: 0.6415 vs. 0.6184, P = .781). Interobserver agreement for ADC was good (ICC 0.83-0.86). Conclusions: Integrating ADC-derived DWI metrics with RECIST 1.1 may provide complementary information for early assessment of pelvic bone metastases from lung cancer, particularly in lesions categorized as stable disease. Advances in knowledge: This pilot study proposes quantitative ADC thresholds and introduces a framework integrating these with RECIST 1.1 to support more objective response assessment in lung cancer pelvic bone metastases.
- 19 October 2026
- Published by Oxford University Press (OUP) in British Journal of Radiology
Integrating diffusion-weighted imaging parameters with RECIST 1.1 for early assessment of systemic therapy response in lung cancer pelvic bone metastases
Objectives: To determine preliminary quantitative ADC thresholds for functional response and to evaluate whether integrating DWI metrics with RECIST 1.1 provides complementary information for early systemic therapy response assessment in bone metastases. Methods: This retrospective single-center study included lung cancer patients with pelvic bone metastases who underwent 3 MRIs between November 2015 and November 2023: baseline, early time point ( 1300 µm2/s or ΔADC > 0.14. An integrated response scheme (DWI-RECIST) combined this DWI-based response with RECIST assessment. Early responses at TP1 were compared between RECIST and DWI-RECIST using RECIST at TP2 as reference standard. Interobserver agreement was assessed using intraclass correlation coefficient, and diagnostic performance was evaluated using net reclassification improvement and AUCs. Results: Seventy-eight lesions from 27 patients were included. Compared with RECIST, DWI-RECIST yielded net classification gains of 7.4% (reader 1) and 11.1% (reader 2). DWI-RECIST showed higher AUCs than RECIST (reader 1: 0.6964 vs. 0.6184, P = .472; reader 2: 0.6415 vs. 0.6184, P = .781). Interobserver agreement for ADC was good (ICC 0.83-0.86). Conclusions: Integrating ADC-derived DWI metrics with RECIST 1.1 may provide complementary information for early assessment of pelvic bone metastases from lung cancer, particularly in lesions categorized as stable disease. Advances in knowledge: This pilot study proposes quantitative ADC thresholds and introduces a framework integrating these with RECIST 1.1 to support more objective response assessment in lung cancer pelvic bone metastases.
- Open Access
- 29 May 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (11) , 1689
Radiograph-Based Deep Learning Model to Support Finger Joint Selection for Ultrasound Examination in Rheumatoid Arthritis
Background/Objectives: Ultrasound is the standard imaging modality to evaluate the inflammatory changes in hand joints of rheumatoid arthritis (RA) patients. However, it is operator-dependent and takes a long time to examine. In this study, we developed a radiograph-based deep learning (DL) model to support prioritization of finger joints for ultrasound (US) examination in RA patients. Methods: In this retrospective study, hand radiographs from RA patients who underwent same-day US examination of bilateral finger joints were analyzed. A DL model was developed using hand radiographs from 270 patients (2043 finger joints) to estimate joint-level likelihood of inflammatory activity. US findings served as the reference standard for model training, while clinical findings of joint tenderness and swelling were incorporated as additional tabular inputs. Model performance was evaluated in a temporal-split test cohort consisting of 40 patients (270 joints) and compared with the performance of a clinical-only logistic regression model based on joint tenderness and swelling. Results: In the test set, the DL model demonstrated higher sensitivity (82.1% vs. 38.5%), negative predictive value (96.8% vs. 90.3%), and F1-score (69.6% vs. 48.4%) than the clinical-only model. Although the area under the receiver operating characteristic curve did not differ significantly between models (p = 0.43), precision–recall (PR) analysis showed superior performance of the DL model, with a higher area under the PR curve (0.625 vs. 0.540). At the threshold maximizing the F1-score, DL-assisted triage reduced the number of finger joints selected for US examination by approximately 80%. Conclusions: A radiograph-based DL model can support efficient prioritization of finger joints for US examination in RA, offering a practical approach to enhance joint-level US triage in routine clinical practice.
- 29 May 2026
- Published by MDPI AG in Diagnostics
- Vol. 16 (11) , 1689
Open AccessRadiograph-Based Deep Learning Model to Support Finger Joint Selection for Ultrasound Examination in Rheumatoid Arthritis
Background/Objectives: Ultrasound is the standard imaging modality to evaluate the inflammatory changes in hand joints of rheumatoid arthritis (RA) patients. However, it is operator-dependent and takes a long time to examine. In this study, we developed a radiograph-based deep learning (DL) model to support prioritization of finger joints for ultrasound (US) examination in RA patients. Methods: In this retrospective study, hand radiographs from RA patients who underwent same-day US examination of bilateral finger joints were analyzed. A DL model was developed using hand radiographs from 270 patients (2043 finger joints) to estimate joint-level likelihood of inflammatory activity. US findings served as the reference standard for model training, while clinical findings of joint tenderness and swelling were incorporated as additional tabular inputs. Model performance was evaluated in a temporal-split test cohort consisting of 40 patients (270 joints) and compared with the performance of a clinical-only logistic regression model based on joint tenderness and swelling. Results: In the test set, the DL model demonstrated higher sensitivity (82.1% vs. 38.5%), negative predictive value (96.8% vs. 90.3%), and F1-score (69.6% vs. 48.4%) than the clinical-only model. Although the area under the receiver operating characteristic curve did not differ significantly between models (p = 0.43), precision–recall (PR) analysis showed superior performance of the DL model, with a higher area under the PR curve (0.625 vs. 0.540). At the threshold maximizing the F1-score, DL-assisted triage reduced the number of finger joints selected for US examination by approximately 80%. Conclusions: A radiograph-based DL model can support efficient prioritization of finger joints for US examination in RA, offering a practical approach to enhance joint-level US triage in routine clinical practice.
- 16 December 2025
- Published by Springer Nature in Skeletal Radiology
- Vol. 55 (5) , 1037-1044
Synthetic T2-weighted images and T2 quantification using GRAPPATINI for shoulder joint evaluation: comparison with conventional T2-weighted images
Objective To evaluate the diagnostic equivalence of GRAPPATINI-derived synthetic fat-saturated (fs-) T2-weighted image (T2WI) compared with conventional fs-T2WI for assessing supraspinatus tendon (SST) and superior labrum, and to investigate the correlation between GRAPPATINI-derived T2 values and glenohumeral joint cartilage degeneration. Materials and Methods This retrospective study included 23 patients (3 men, 20 women; median age, 62 years) who underwent 3.0 T shoulder MRI including both conventional and synthetic fs-T2WIs. Image quality was quantitatively assessed using relative signal-to-noise ratio (rSNR) and contrast-to-noise ratio (rCNR) at fluid, muscle, glenoid cartilage, humeral cartilage, and tendon, and compared using Wilcoxon signed rank test. Four radiologists independently evaluated the detectability of SST and superior labral lesions, and diagnostic equivalence was assessed using an interchangeability analysis. T2 values were measured in two regions of interest in each of the glenoid and humeral head cartilage, and compared between degenerated and non-degenerated cartilage on conventional T2WI using Mann–Whitney U test. Results rSNR differed significantly in glenoid cartilage (p = 0.037) and tendon (p = 0.007), whereas rCNR showed no significant differences. For lesion detectability, conventional and synthetic fs-T2WI were interchangeable in evaluating SST and superior labral abnormalities (upper limits of 95% confidence interval for differences in agreement rates ≤ 3.7% for SST and ≤ 2.7% for labrum). T2 values were significantly higher in degenerated cartilage than in non-degenerated cartilage in both the glenoid (p = 0.01) and the humeral head (p = 0.02). Conclusion GRAPPATINI-derived synthetic fs-T2WIs were diagnostically equivalent to conventional fs-T2WI for evaluating SST and superior labral lesions. In addition, GRAPPATINI-derived T2 maps enabled supplementary quantitative assessment of cartilage degeneration.
- 16 December 2025
- Published by Springer Nature in Skeletal Radiology
- Vol. 55 (5) , 1037-1044
Synthetic T2-weighted images and T2 quantification using GRAPPATINI for shoulder joint evaluation: comparison with conventional T2-weighted images
Objective To evaluate the diagnostic equivalence of GRAPPATINI-derived synthetic fat-saturated (fs-) T2-weighted image (T2WI) compared with conventional fs-T2WI for assessing supraspinatus tendon (SST) and superior labrum, and to investigate the correlation between GRAPPATINI-derived T2 values and glenohumeral joint cartilage degeneration. Materials and Methods This retrospective study included 23 patients (3 men, 20 women; median age, 62 years) who underwent 3.0 T shoulder MRI including both conventional and synthetic fs-T2WIs. Image quality was quantitatively assessed using relative signal-to-noise ratio (rSNR) and contrast-to-noise ratio (rCNR) at fluid, muscle, glenoid cartilage, humeral cartilage, and tendon, and compared using Wilcoxon signed rank test. Four radiologists independently evaluated the detectability of SST and superior labral lesions, and diagnostic equivalence was assessed using an interchangeability analysis. T2 values were measured in two regions of interest in each of the glenoid and humeral head cartilage, and compared between degenerated and non-degenerated cartilage on conventional T2WI using Mann–Whitney U test. Results rSNR differed significantly in glenoid cartilage (p = 0.037) and tendon (p = 0.007), whereas rCNR showed no significant differences. For lesion detectability, conventional and synthetic fs-T2WI were interchangeable in evaluating SST and superior labral abnormalities (upper limits of 95% confidence interval for differences in agreement rates ≤ 3.7% for SST and ≤ 2.7% for labrum). T2 values were significantly higher in degenerated cartilage than in non-degenerated cartilage in both the glenoid (p = 0.01) and the humeral head (p = 0.02). Conclusion GRAPPATINI-derived synthetic fs-T2WIs were diagnostically equivalent to conventional fs-T2WI for evaluating SST and superior labral lesions. In addition, GRAPPATINI-derived T2 maps enabled supplementary quantitative assessment of cartilage degeneration.
- 3 November 2025
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- Vol. 39 (4) , 1-14
Patch-Wise Approach with Vision Transformer for Detecting Implant Failure in Spinal Radiography
Spinal implant failures are becoming more common due to the increasing number of surgeries and an aging population. Early detection is crucial but often missed on radiographs due to subtle features and limited radiologist availability. This study aims to develop and validate a vision transformer-based deep learning model for detecting spinal implant fractures and to assess its impact on radiologist performance. The training and testing datasets included 9924 spinal radiographs (3492 studies) from 798 patients, between 2003 and 2023. A DINOv2-based model first detected spinal implants in radiographs, then analyzed 224 × 224-pixel patches within those regions to classify the presence or absence of implant fractures. Three radiologists (1 to 16 years of experience) independently reviewed the test set (1538 images) with and without AI assistance. Performance was measured using accuracy, F1 score, precision, recall, and generalized estimating equation (GEE) analysis. The AI model achieved a recall of 0.94, precision of 0.37, F1 score of 0.54, and accuracy of 0.83. With AI assistance, radiologist recall improved from 0.70 to 0.95, with the greatest gain seen in the least experienced reader (0.49 to 0.92). GEE analysis confirmed significant diagnostic improvement (OR = 2.82, p < 0.001), particularly in rod fractures. This patch-wise transformer-based approach demonstrated high sensitivity and improved reader performance, supporting its use as a triage tool in spinal implant surveillance.
- 3 November 2025
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- Vol. 39 (4) , 1-14
Patch-Wise Approach with Vision Transformer for Detecting Implant Failure in Spinal Radiography
Spinal implant failures are becoming more common due to the increasing number of surgeries and an aging population. Early detection is crucial but often missed on radiographs due to subtle features and limited radiologist availability. This study aims to develop and validate a vision transformer-based deep learning model for detecting spinal implant fractures and to assess its impact on radiologist performance. The training and testing datasets included 9924 spinal radiographs (3492 studies) from 798 patients, between 2003 and 2023. A DINOv2-based model first detected spinal implants in radiographs, then analyzed 224 × 224-pixel patches within those regions to classify the presence or absence of implant fractures. Three radiologists (1 to 16 years of experience) independently reviewed the test set (1538 images) with and without AI assistance. Performance was measured using accuracy, F1 score, precision, recall, and generalized estimating equation (GEE) analysis. The AI model achieved a recall of 0.94, precision of 0.37, F1 score of 0.54, and accuracy of 0.83. With AI assistance, radiologist recall improved from 0.70 to 0.95, with the greatest gain seen in the least experienced reader (0.49 to 0.92). GEE analysis confirmed significant diagnostic improvement (OR = 2.82, p < 0.001), particularly in rod fractures. This patch-wise transformer-based approach demonstrated high sensitivity and improved reader performance, supporting its use as a triage tool in spinal implant surveillance.
- Open Access
- 21 May 2025
- Published by Springer Nature in BMC Medical Imaging
- Vol. 25 (1) , 1-9
Evaluating the diagnostic accuracy of dynamic CT transmural sign for T staging of gastric cancer compared to conventional CT criteria
Background: To retrospectively evaluate the diagnostic performance of pre-procedural multidetector computed tomography (MDCT) using dynamic CT transmural (CTTM) criteria for T staging of gastric cancer. Methods: This retrospective study enrolled 116 patients who underwent three-phase dynamic MDCT and subsequently received endoscopic treatment or surgery. The positive CTTM sign was defined as a new diagnostic criterion for this study. Two radiologists independently reviewed the CT images, categorizing patients into five groups: T1b sm1, T1b sm2/3, T2, T3, and T4. The diagnostic performance of the new criterion was assessed against pathological results as the gold standard. Sensitivity, specificity, accuracy, and positive and negative predictive values for each T stage were compared to those of conventional CT. Results: Dynamic CTTM criteria demonstrated higher overall diagnostic accuracy (Reviewer 1: 88.9%, Reviewer 2: 91.4%) for T staging compared to conventional CT criteria (75.9%). Reviewer 1 accurately staged T1b in 85.3% of patients, T2 in 87.7%, T3 in 94.1%, and T4 in 98.3%. Reviewer 2 achieved T1b accuracy of 87.9%, T2 of 89.2%, T3 of 94.9%, and T4 of 98.3%. The dynamic CTTM criteria effectively subdivided T1b into T1b sm1 (87.2% and 88.1%) and T1b sm2/3 (84.6% and 86.2%). Dynamic CTTM criteria exhibited higher under-staging rates, while conventional criteria showed higher over-staging rates. Conclusions: Dynamic CTTM criteria demonstrated superior accuracy in T staging of gastric cancer and reasonable effectiveness in subdividing T1b stages into T1b sm1 and T1b sm2/3.
- 21 May 2025
- Published by Springer Nature in BMC Medical Imaging
- Vol. 25 (1) , 1-9
Open AccessEvaluating the diagnostic accuracy of dynamic CT transmural sign for T staging of gastric cancer compared to conventional CT criteria
Background: To retrospectively evaluate the diagnostic performance of pre-procedural multidetector computed tomography (MDCT) using dynamic CT transmural (CTTM) criteria for T staging of gastric cancer. Methods: This retrospective study enrolled 116 patients who underwent three-phase dynamic MDCT and subsequently received endoscopic treatment or surgery. The positive CTTM sign was defined as a new diagnostic criterion for this study. Two radiologists independently reviewed the CT images, categorizing patients into five groups: T1b sm1, T1b sm2/3, T2, T3, and T4. The diagnostic performance of the new criterion was assessed against pathological results as the gold standard. Sensitivity, specificity, accuracy, and positive and negative predictive values for each T stage were compared to those of conventional CT. Results: Dynamic CTTM criteria demonstrated higher overall diagnostic accuracy (Reviewer 1: 88.9%, Reviewer 2: 91.4%) for T staging compared to conventional CT criteria (75.9%). Reviewer 1 accurately staged T1b in 85.3% of patients, T2 in 87.7%, T3 in 94.1%, and T4 in 98.3%. Reviewer 2 achieved T1b accuracy of 87.9%, T2 of 89.2%, T3 of 94.9%, and T4 of 98.3%. The dynamic CTTM criteria effectively subdivided T1b into T1b sm1 (87.2% and 88.1%) and T1b sm2/3 (84.6% and 86.2%). Dynamic CTTM criteria exhibited higher under-staging rates, while conventional criteria showed higher over-staging rates. Conclusions: Dynamic CTTM criteria demonstrated superior accuracy in T staging of gastric cancer and reasonable effectiveness in subdividing T1b stages into T1b sm1 and T1b sm2/3.
- 1 September 2023
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 308 (3) , e230667
Development of a Semiquantitative Whole-Body MRI Scoring System for Multiple Myeloma
Background In patients with multiple myeloma (MM), the serum marker β2-microglobulin does not always accurately reflect tumor load. In contrast, whole-body (WB) MRI has shown high sensitivity for detecting bone lesions. Purpose To develop and validate a semiquantitative WB MRI scoring system for newly diagnosed MM and to compare it with the International Staging System (ISS) and Revised ISS (R-ISS). Materials and Methods This study included two retrospective groups (group 1, July 2015 to September 2021; group 2, February 2020 to September 2021) and one prospective group (group 3, October 2021 to February 2022) of patients with newly diagnosed MM. A new scoring system for MM was developed using spine MRI scans in group 1 and WB MRI scans in group 2 that integrated three features: (a) background marrow pattern, (b) number of focal bone lesions, and (c) presence of extramedullary or paramedullary lesions. The summed total score ranged from zero to nine. The interobserver agreement for each feature was assessed using Fleiss or Cohen weighted κ. WB MRI total scores in group 3 were compared across ISS and R-ISS stages using two-way analysis of variance. Results Groups 1, 2, and 3 included 103 patients (mean age, 62.1 years ± 9.1 [SD]; 60 men), 36 patients (mean age 65.4 years ± 11.3 [SD]; 19 women), and 39 participants (mean age, 62.0 years ± 11.7 [SD]; 20 men), respectively. The interobserver agreements for the three features composing the scoring system were substantial (κ range, 0.69-0.80). WB MRI total score increased with increasing ISS stage (mean score for ISS 1, 2, and 3 was 2.2, 4.2, and 5.8, respectively; P = .009) and R-ISS stage (mean score for R-ISS 1, 2, and 3 was 2.1, 3.8, and 5.9, respectively; P = .005). Conclusion The developed WB MRI scoring system for MM demonstrated substantial observer agreement and corresponded well with ISS and R-ISS stages. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Dragan and Messiou in this issue.
- 1 September 2023
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 308 (3) , e230667
Development of a Semiquantitative Whole-Body MRI Scoring System for Multiple Myeloma
Background In patients with multiple myeloma (MM), the serum marker β2-microglobulin does not always accurately reflect tumor load. In contrast, whole-body (WB) MRI has shown high sensitivity for detecting bone lesions. Purpose To develop and validate a semiquantitative WB MRI scoring system for newly diagnosed MM and to compare it with the International Staging System (ISS) and Revised ISS (R-ISS). Materials and Methods This study included two retrospective groups (group 1, July 2015 to September 2021; group 2, February 2020 to September 2021) and one prospective group (group 3, October 2021 to February 2022) of patients with newly diagnosed MM. A new scoring system for MM was developed using spine MRI scans in group 1 and WB MRI scans in group 2 that integrated three features: (a) background marrow pattern, (b) number of focal bone lesions, and (c) presence of extramedullary or paramedullary lesions. The summed total score ranged from zero to nine. The interobserver agreement for each feature was assessed using Fleiss or Cohen weighted κ. WB MRI total scores in group 3 were compared across ISS and R-ISS stages using two-way analysis of variance. Results Groups 1, 2, and 3 included 103 patients (mean age, 62.1 years ± 9.1 [SD]; 60 men), 36 patients (mean age 65.4 years ± 11.3 [SD]; 19 women), and 39 participants (mean age, 62.0 years ± 11.7 [SD]; 20 men), respectively. The interobserver agreements for the three features composing the scoring system were substantial (κ range, 0.69-0.80). WB MRI total score increased with increasing ISS stage (mean score for ISS 1, 2, and 3 was 2.2, 4.2, and 5.8, respectively; P = .009) and R-ISS stage (mean score for R-ISS 1, 2, and 3 was 2.1, 3.8, and 5.9, respectively; P = .005). Conclusion The developed WB MRI scoring system for MM demonstrated substantial observer agreement and corresponded well with ISS and R-ISS stages. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Dragan and Messiou in this issue.
- Open Access
- 29 June 2023
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 12 (13) , 4369
Diagnosis of Local Recurrence of Malignant Soft Tissue Tumors after Reconstructive Surgery on MRI
Purpose: Magnetic resonance imaging (MRI) is useful in the diagnosis of local recurrence, but few studies have explored recurrence in MRI in patients after reconstructive surgery. The purpose of this study was to analyze MRI findings of locoregional recurrence following reconstructive surgery after malignant soft tissue tumor resection. Method: Fifty-three postoperative MRIs from 37 patients who underwent reconstructive surgery after malignant soft tissue tumor resection were retrospectively reviewed. A total of 76 enhancing lesions, including 40 locoregional recurrences and 36 postoperative changes, were analyzed regarding morphology (location on the transplanted tissue, border, and shape) and the signals on T1- and T2-weighted imaging (T1WI, T2WI), fat-suppressed (FS) T2WI, and contrast-enhanced FS T1WI. Diffusion-weighted imaging with an apparent diffusion coefficient was assessed. A chi-squared test and Fisher’s exact test were used for statistical analysis. Results: The most common site of recurrent tumors and postoperative changes was the peripheral margin on transplanted tissue (63% and 61%, respectively p = 0.907). Recurrent tumors commonly appeared with well-defined borders (75%) as well as nodular appearance (98%), hyperintensity on T2WI (85%) and FS-T2WI (95%), isointensity on T1WI (65%), impeded water diffusion (55%), and intense (50%) or moderate (45%) enhancement. Postoperative changes showed ill-defined borders (75%), nodular appearance (56%), facilitated water diffusion (69%), and moderate (86%) enhancement, which were significantly different from those of recurrent tumors (p ≤ 0.020). Conclusions: Common and partitioning MRI features of locoregional recurrence were well-defined borders, nodular shape, impeded water diffusion, and intense enhancement. Peripheral margins on transplanted tissue were common sites in both recurrent tumors and postoperative changes.
- 29 June 2023
- Published by MDPI AG in Journal of Clinical Medicine
- Vol. 12 (13) , 4369
Open AccessDiagnosis of Local Recurrence of Malignant Soft Tissue Tumors after Reconstructive Surgery on MRI
Purpose: Magnetic resonance imaging (MRI) is useful in the diagnosis of local recurrence, but few studies have explored recurrence in MRI in patients after reconstructive surgery. The purpose of this study was to analyze MRI findings of locoregional recurrence following reconstructive surgery after malignant soft tissue tumor resection. Method: Fifty-three postoperative MRIs from 37 patients who underwent reconstructive surgery after malignant soft tissue tumor resection were retrospectively reviewed. A total of 76 enhancing lesions, including 40 locoregional recurrences and 36 postoperative changes, were analyzed regarding morphology (location on the transplanted tissue, border, and shape) and the signals on T1- and T2-weighted imaging (T1WI, T2WI), fat-suppressed (FS) T2WI, and contrast-enhanced FS T1WI. Diffusion-weighted imaging with an apparent diffusion coefficient was assessed. A chi-squared test and Fisher’s exact test were used for statistical analysis. Results: The most common site of recurrent tumors and postoperative changes was the peripheral margin on transplanted tissue (63% and 61%, respectively p = 0.907). Recurrent tumors commonly appeared with well-defined borders (75%) as well as nodular appearance (98%), hyperintensity on T2WI (85%) and FS-T2WI (95%), isointensity on T1WI (65%), impeded water diffusion (55%), and intense (50%) or moderate (45%) enhancement. Postoperative changes showed ill-defined borders (75%), nodular appearance (56%), facilitated water diffusion (69%), and moderate (86%) enhancement, which were significantly different from those of recurrent tumors (p ≤ 0.020). Conclusions: Common and partitioning MRI features of locoregional recurrence were well-defined borders, nodular shape, impeded water diffusion, and intense enhancement. Peripheral margins on transplanted tissue were common sites in both recurrent tumors and postoperative changes.
- Open Access
- 16 February 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (4) , 755
Differentiating Multiple Myeloma and Osteolytic Bone Metastases on Contrast-Enhanced Computed Tomography Scans: The Feasibility of Radiomics Analysis
Osteolytic lesions can be seen in both multiple myeloma (MM), and osteolytic bone metastasis on computed tomography (CT) scans. We sought to assess the feasibility of a CT-based radiomics model to distinguish MM from metastasis. This study retrospectively included patients with pre-treatment thoracic or abdominal contrast-enhanced CT from institution 1 (training set: 175 patients with 425 lesions) and institution 2 (external test set: 50 patients with 85 lesions). After segmenting osteolytic lesions on CT images, 1218 radiomics features were extracted. A random forest (RF) classifier was used to build the radiomics model with 10-fold cross-validation. Three radiologists distinguished MM from metastasis using a five-point scale, both with and without the assistance of RF model results. Diagnostic performance was evaluated using the area under the curve (AUC). The AUC of the RF model was 0.807 and 0.762 for the training and test set, respectively. The AUC of the RF model and the radiologists (0.653–0.778) was not significantly different for the test set (p ≥ 0.179). The AUC of all radiologists was significantly increased (0.833–0.900) when they were assisted by RF model results (p < 0.001). In conclusion, the CT-based radiomics model can differentiate MM from osteolytic bone metastasis and improve radiologists’ diagnostic performance.
- 16 February 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (4) , 755
Open AccessDifferentiating Multiple Myeloma and Osteolytic Bone Metastases on Contrast-Enhanced Computed Tomography Scans: The Feasibility of Radiomics Analysis
Osteolytic lesions can be seen in both multiple myeloma (MM), and osteolytic bone metastasis on computed tomography (CT) scans. We sought to assess the feasibility of a CT-based radiomics model to distinguish MM from metastasis. This study retrospectively included patients with pre-treatment thoracic or abdominal contrast-enhanced CT from institution 1 (training set: 175 patients with 425 lesions) and institution 2 (external test set: 50 patients with 85 lesions). After segmenting osteolytic lesions on CT images, 1218 radiomics features were extracted. A random forest (RF) classifier was used to build the radiomics model with 10-fold cross-validation. Three radiologists distinguished MM from metastasis using a five-point scale, both with and without the assistance of RF model results. Diagnostic performance was evaluated using the area under the curve (AUC). The AUC of the RF model was 0.807 and 0.762 for the training and test set, respectively. The AUC of the RF model and the radiologists (0.653–0.778) was not significantly different for the test set (p ≥ 0.179). The AUC of all radiologists was significantly increased (0.833–0.900) when they were assisted by RF model results (p < 0.001). In conclusion, the CT-based radiomics model can differentiate MM from osteolytic bone metastasis and improve radiologists’ diagnostic performance.
- Open Access
- 1 January 2023
- Published by XMLink in Investigative Magnetic Resonance Imaging
- Vol. 27 (2) , 75-83
Diagnosis of Infectious Spondylitis Using Non-Contrast Enhanced MRI With Axial Diffusion-Weighted Images: Comparison With Gadolinium-Enhanced MRI
Choi SH, et al. Investig Magn Reson Imaging. 2023 Jun;27(2):75-83. https://doi.org/10.13104/imri.2022.1105
- 1 January 2023
- Published by XMLink in Investigative Magnetic Resonance Imaging
- Vol. 27 (2) , 75-83
Open AccessDiagnosis of Infectious Spondylitis Using Non-Contrast Enhanced MRI With Axial Diffusion-Weighted Images: Comparison With Gadolinium-Enhanced MRI
Choi SH, et al. Investig Magn Reson Imaging. 2023 Jun;27(2):75-83. https://doi.org/10.13104/imri.2022.1105
- 1 September 2022
- Published by Oxford University Press (OUP) in British Journal of Radiology
- Vol. 95 (1137) , 20220009
Differentiation of multiple myelomas from osteolytic bone metastases: Diagnostic value of tumor homogeneity on Contrast-Enhanced CT
Objectives: To investigate the diagnostic value of tumor homogeneity on contrast-enhanced (CE) computed tomography (CT) to differentiate multiple myeloma (MM) from osteolytic bone metastases (Mets). Methods: This retrospective study included patients who were diagnosed with MM or Mets and had multiple (≥2) osteolytic bone tumors on pre-treatment CE-CT. Intratumoral homogeneity was assessed by coefficient of variation (CV, ratio of standard deviation to mean) of the density of a single lesion (CV-lesion). Intertumoral homogeneity was assessed as the CV of the densities of multiple lesions in one patient (CV-patient). A classification model was built from CT parameters using classification and regression tree (CART) analysis. Diagnostic performance of the model was evaluated using C-statistics. Results: A total of 272 lesions (81 MM and 191 Mets) of 105 patients were analyzed. The mean CV-lesion and CV-patient of MM were significantly lower than those of Mets: 0.17 vs 0.26 for CV-lesion (p = 0.005) and 0.16 vs 0.23 for CV-patient (p = 0.013). Thickened struts were more common in MM than in Mets (49.1% vs 12.8%, p ≤ 0.001). In CART analysis, CV-lesion was the first partitioning predictor, followed by thickened struts and by CV patient. The CART model could distinguish MM from Mets in both the model development cohort (C-statistic: 0.843) and the temporal validation cohort (0.721, 0.686, and 0.686 for three reviewers, respectively). Conclusions: MM showed intratumoral and intertumoral homogeneity compared with Mets on CE-CT. The combination of CV-lesion and CV-patient can be helpful to radiologists in differentiation of MM from Mets. Advances in knowledge: Our study showed that MM had intratumoral and intertumoral homogeneity compared with Mets on contrast-enhanced CT.
- 1 September 2022
- Published by Oxford University Press (OUP) in British Journal of Radiology
- Vol. 95 (1137) , 20220009
Differentiation of multiple myelomas from osteolytic bone metastases: Diagnostic value of tumor homogeneity on Contrast-Enhanced CT
Objectives: To investigate the diagnostic value of tumor homogeneity on contrast-enhanced (CE) computed tomography (CT) to differentiate multiple myeloma (MM) from osteolytic bone metastases (Mets). Methods: This retrospective study included patients who were diagnosed with MM or Mets and had multiple (≥2) osteolytic bone tumors on pre-treatment CE-CT. Intratumoral homogeneity was assessed by coefficient of variation (CV, ratio of standard deviation to mean) of the density of a single lesion (CV-lesion). Intertumoral homogeneity was assessed as the CV of the densities of multiple lesions in one patient (CV-patient). A classification model was built from CT parameters using classification and regression tree (CART) analysis. Diagnostic performance of the model was evaluated using C-statistics. Results: A total of 272 lesions (81 MM and 191 Mets) of 105 patients were analyzed. The mean CV-lesion and CV-patient of MM were significantly lower than those of Mets: 0.17 vs 0.26 for CV-lesion (p = 0.005) and 0.16 vs 0.23 for CV-patient (p = 0.013). Thickened struts were more common in MM than in Mets (49.1% vs 12.8%, p ≤ 0.001). In CART analysis, CV-lesion was the first partitioning predictor, followed by thickened struts and by CV patient. The CART model could distinguish MM from Mets in both the model development cohort (C-statistic: 0.843) and the temporal validation cohort (0.721, 0.686, and 0.686 for three reviewers, respectively). Conclusions: MM showed intratumoral and intertumoral homogeneity compared with Mets on CE-CT. The combination of CV-lesion and CV-patient can be helpful to radiologists in differentiation of MM from Mets. Advances in knowledge: Our study showed that MM had intratumoral and intertumoral homogeneity compared with Mets on contrast-enhanced CT.