Tag: pubmed

Quant Imaging Med Surg. 2024 Dec 5;14(12):9323-9334. doi: 10.21037/qims-24-1230. Epub 2024 Nov 29.

ABSTRACT

BACKGROUND: Quantitative assessment of cortical bone microstructure is crucial for the evaluation of osteoporosis, yet current clinical methods such as dual-energy X-ray absorptiometry (DXA) have many limitations. The in vivo quantitative assessment without radiation can be achieved by ultrashort echo time (UTE) magnetic resonance imaging (MRI), where double-echo UTE has high clinical feasibility. However, related studies have mainly focused on distal extremities, and there is a lack of in vivo studies on the proximal femur. This cross-sectional study, as a preliminary study for in vivo quantitative assessment of proximal femoral cortical bone in healthy adults by double-echo UTE MRI, aimed to evaluate the repeatability and explore the impact of potential influencing factors on UTE measurements, thereby providing a reference for the further clinical application of this technique.

METHODS: Healthy volunteers without osteoporosis risk factors were enrolled and underwent double-echo UTE MRI of the proximal femur. Porosity index (PI) and cortical bone thickness (CbTh) were obtained manually by two radiologists independently on double-echo UTE images using image processing software. Repeatability of PI and CbTh measurements were evaluated using intra-class correlation coefficient (ICC) analysis. PI and CbTh of different limbs and sexes were compared, and P<0.05 indicated statistical significance for these analyses. Correlations of PI and CbTh with age and body mass index (BMI) were assessed by Pearson or Spearman correlation coefficient as well as the partial Pearson correlation coefficient. Curve estimation was used to assess non-monotonic variable correlations in scatterplots. For these analyses, P<0.025 indicated statistical significance according to the Bonferroni correction.

RESULTS: A total of 52 healthy cases (33 males, 19 females) aged 22-55 years were included for analysis, where no statistical difference in age was found between sexes (P=0.586) and BMI of males was greater than that of females (P=0.007). The repeatability of PI and CbTh measurements was excellent (ICC 0.985 and 0.943, respectively). The proximal femoral cortical PI was greater on the non-dominant side (P<0.001). PI was greater in males than in females (non-dominant: P<0.001, dominant: P=0.032) and CbTh was greater on the non-dominant side in males than in females (P=0.036) after excluding the effect of confounding factor (BMI). PI on the dominant side was positively correlated with BMI in males (r=0.535, P=0.001), and CbTh on the non-dominant side was positively correlated with BMI in males (r=0.482, P=0.005). There was a U-shaped curve trend between dominant side cortical PI and age in females (y = 65.32 – 1.88x + 0.02x², R²=0.348, P=0.033), although there was no statistical significance after Bonferroni correction.

CONCLUSIONS: Double-echo UTE MRI enables in vivo quantitative assessment of proximal femoral cortical microarchitecture with excellent repeatability. We identified the effects of limb dominance, sex, age, and BMI on UTE measurements in healthy adults, which can serve as a reference for future in vivo studies on proximal femoral cortical bone and is essential before clinical application.

PMID:39698732 | PMC:PMC11652011 | DOI:10.21037/qims-24-1230

Quant Imaging Med Surg. 2024 Dec 5;14(12):8601-8613. doi: 10.21037/qims-24-1035. Epub 2024 Nov 24.

ABSTRACT

BACKGROUND: Currently, the activity of juvenile dermatomyositis (JDM) is mainly assessed based on clinical manifestations, creatine kinase (CK) level, and magnetic resonance imaging (MRI), but certain limitations arise in these approaches for children. Thus, this cross-sectional study aimed to explore the value of ultrasound in evaluating muscle inflammation via the dynamic analysis of muscle ultrasound characteristics in children with active or stable JDM.

METHODS: The data of a group of children who were diagnosed with JDM and admitted to the Rheumatology and Immunology Department of the Capital Pediatric Research Institute Children’s Hospital between June 2022 and November 2023, and a normal control group were collected. The clinical, ultrasound, and laboratory data of the children with active and stable JDM were collected and compared with those of a normal control group. Muscle thickness (MT), fascia thickness (FT), muscle echo intensity (EI), muscle microvascular flow imaging (MVFI) distribution, and the blood flow resistance index (RI) were measured via ultrasound for the comparative analysis. A Spearman correlation analysis was conducted to assess the correlation between the ultrasound parameters in the JDM patients, and muscle scores and laboratory indicators of disease activity. Receiver operating characteristic (ROC) curves were generated for the ultrasound parameters. Five active JDM children were dynamically followed up.

RESULTS: The data of 26 children with active JDM, 29 with stable JDM, and 31 normal control children were collected. The patients with active JDM had significantly greater EI (median 68.9 vs. 47.4, P<0.01), increased FT (median 0.25 vs. 0.15, P<0.01), and an increased MVFI distribution (P<0.01). The Spearman correlation analysis revealed a negative correlation between increased EI, FT, and MVFI distribution, and the Childhood Myositis Assessment Scale (CMAS) score (R=-0.662, -0.673, -0.667, all P<0.05). There was a statistically significant difference in EI between the stable JDM children and healthy children (median 47.4 vs. 39.0, P<0.05). During the follow-up period, two children with slow fasciitis resolution developed soft tissue calcification.

CONCLUSIONS: EI, FT, and MVFI distribution can be used to assess the activity status of individuals with JDM, and are correlated with clinical activity indices. EI may be abnormal even in children with a stable clinical condition, which suggests that ultrasound may more accurately reflect muscle status. Children with long-term unresolved fasciitis may be at risk of developing calcification.

PMID:39698728 | PMC:PMC11651955 | DOI:10.21037/qims-24-1035

Quant Imaging Med Surg. 2024 Dec 5;14(12):8520-8537. doi: 10.21037/qims-24-941. Epub 2024 Oct 24.

ABSTRACT

BACKGROUND: Given the limited capacity and suboptimal sensitivity of ultrasonography (US), gadoxetic acid disodium (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) demonstrates good diagnostic performance for hepatocellular carcinoma (HCC). Some researchers have proposed that the abbreviated MRI (AMRI) protocols have potential as a surveillance tool. However, few studies have compared multiple AMRI protocols with complete Gd-EOB-DTPA contrast-enhanced MRI for HCC surveillance. We aimed to explore and compare the diagnostic performance of 3 AMRI protocols as HCC surveillance in high-risk patients.

METHODS: This multi-center, retrospective, blinded reader study conducted in China consecutively enrolled 339 patients with hepatitis and/or cirrhosis who underwent complete Gd-EOB-DTPA contrast-enhanced MRI for HCC surveillance from 2020 to 2023. We extracted 3 additional AMRI protocols: noncontrast-AMRI [NC-AMRI: T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI)]; dynamic-AMRI (Dyn-AMRI: early and late arterial phases, portal venous phase, and DWI); and hepatobiliary phase-AMRI (HBP-AMRI: T2WI, DWI, and HBP). Then, 2 independent radiologists assessed the AMRI and complete Gd-EOB-DTPA contrast-enhanced MRI protocols. Patients were classified as HCC positive/HCC negative based on the reference standard. Agreement was assessed using Kappa statistics. The acquisition time differences of the 4 MRI protocols were analyzed by analysis of variance (ANOVA). Per-lesion HCC diagnostic performances were compared by Cochran’s Q test. Receiver operating characteristic (ROC) curves for the 3 AMRI protocols were evaluated, and the area under the ROC curve (AUROC) was calculated and compared by DeLong’s test.

RESULTS: A total of 353 lesions were detected in the 339 included patients, and 21/339 patients were diagnosed with HCC (prevalence, 6.2%). The inter-observer agreement was good for all 4 MRI protocols (k>0.75). Acquisition times differed significantly (P<0.001), from the shortest to the longest: NC-AMRI (263.44±5.05 s) < HBP-AMRI (269.18±4.93 s) < Dyn-AMRI (307.71±4.93 s) < complete Gd-EOB-DTPA contrast-enhanced MRI (582.03±3.59 s). The sensitivity (Cochran’s Q=14.667, P=0.002) and specificity (Cochran’s Q=59.682, P<0.001) of 4 MRI protocols were statistically significant. HBP-AMRI showed the highest sensitivity (84.00%), whereas Dyn-AMRI exhibited the highest specificity (99.39%) among 3 AMRI protocols. The per-lesion positive predictive value (PPV) for the NC-AMRI, Dyn-AMRI, and HBP-AMRI was 41.66%, 88.89%, and 47.72%, the corresponding negative predictive value (NPV) was 96.21%, 97.31%, and 98.70%, and the number needed to diagnose (NND) for the NC-AMRI, Dyn-AMRI, HBP-AMRI, and complete Gd-EOB-DTPA contrast-enhanced MRI was: 1.865, 1.577, 1.234, and 1.569, respectively. DeLong’s test showed the AUROC value of either Dyn-AMRI or HBP-AMRI was significantly higher than that of NC-AMRI (Z=2.330, P=0.019; Z=2.680, P=0.007, respectively), but no significant difference between HBP-AMRI and Dyn-AMRI (Z=1.643, P=0.100).

CONCLUSIONS: AMRI protocols can be implemented in clinical practice as a patient-centered and tailored regimen for HCC surveillance in China. NC-AMRI might become an optional tool due to its minimal scanning time, lower cost, and exemption from contrast agents. Dyn-AMRI, achieving the highest specificity, is a reliable surveillance strategy. HBP-AMRI as a favorable alternative showed a high sensitivity and NPV while maintaining considerable specificity and NND.

PMID:39698720 | PMC:PMC11651963 | DOI:10.21037/qims-24-941

Quant Imaging Med Surg. 2024 Dec 5;14(12):9220-9233. doi: 10.21037/qims-24-318. Epub 2024 Nov 14.

ABSTRACT

BACKGROUND: Gut microbiota are associated with brain imaging-derived phenotypes (IDPs); however, the specific causal relationship between the gut microbiota and brain iron-related IDPs remains unclear. Thus, we sought to analyze the potential causal effects of gut microbiota on brain iron-related IDPs using Mendelian randomization (MR).

METHODS: We obtained the data of 196 gut microbiota from a genome-wide association study (GWAS) from the MiBioGen database, as well as the data of 18 quantitative susceptibility mapping (QSM) IDPs and 10 T2* IDPs from the United Kingdom Biobank (UKB). We then conducted one-way two-sample MR analyses to examine their causal interactions. To guarantee the robustness of the results, we performed two independent analysis processes by selecting statistically significant instrumental variables (IVs) with a distinct level of statistical strictness, and derived the intersection of these two analyses.

RESULTS: Our results showed that the genus Howardella was positively correlated with the median susceptibility in the right caudate [β: 0.0935, 95% confidence interval (CI): 0.0601, 0.1269, P_{inverse variance weighting (IVW)}=4.00E-08]; the genus Dialister was positively correlated with the median susceptibility in the right accumbens (β: 0.0949, 95% CI: 0.0575, 0.1324, P_IVW=6.90E-07); the genus Butyricicoccus was positively associated with the median T2* in the left hippocampus with the additional deconfounding of the background field gradient (β: 0.1543, 95% CI: 0.0959, 0.2127, P_IVW=2.20E-07); the genus Desulfovibrio was positively related to the T2* white matter hyperintensity (WMH) IDP with WMH volume regressed out (β: 0.1168, 95% CI: 0.0697, 0.1639, P_IVW=1.20E-06). Notably, both the family Defluviitaleaceae (β: -0.1215, 95% CI: -0.1604, -0.0827, P_IVW=8.40E-10) and genus DefluviitaleaceaeUCG011 (β: -0.1142, 95% CI: -0.1614, -0.0670, P_IVW=2.10E-06) were negatively correlated with the median T2* in the right accumbens with the additional deconfounding of the background field gradient.

CONCLUSIONS: This study found genetic evidence that gut microbiota dysbiosis has causal effects on brain iron-related IDPs. Our findings provide novel insights into the diagnosis and therapeutic management of central nervous system (CNS) diseases.

PMID:39698716 | PMC:PMC11652028 | DOI:10.21037/qims-24-318

Quant Imaging Med Surg. 2024 Dec 5;14(12):8629-8643. doi: 10.21037/qims-24-490. Epub 2024 Nov 11.

ABSTRACT

BACKGROUND: Anal fistula is a common anorectal disorder that significantly diminishes the quality of life for affected patients. Accurate preoperative evaluation of the fistula’s traits is essential for customizing surgical strategies, improving patient outcomes, and reducing the likelihood of the disease returning. This study aimed to evaluate the diagnostic accuracy of multi-phase contrast-enhanced fat-suppressed T1-weighted imaging using three-dimensional gradient echo sequence volumetric interpolated breath-hold examination (CE-FS-T1-3D-VIBE) and fat-suppressed T2-weighted imaging combined with diffusion-weighted imaging (FS-T2WI-DWI) sequence in delineating the characteristics of anal fistulas.

METHODS: A case-control study of 168 patients with anal fistula was conducted through the picture archiving and communication systems (PACS; diagnostic imaging workstation). Imaging evaluations were performed using both multi-phase CE-FS-T1-3D-VIBE and FS-T2WI-DWI imaging on a Siemens 3.0T magnetic resonance imaging system (Skyra 3.0T superconducting type). The efficacy of each imaging modality in depicting the clarity, number, and positioning of the internal openings, as well as the identification of primary and secondary fistulas and abscesses, was independently evaluated in a blinded manner by two senior diagnostic radiologists, each with over a decade of experience. Statistical analyses were performed using χ² test.

RESULTS: Comparative analysis of the FS-T2WI-DWI and multi-phase CE-FS-T1-3D-VIBE sequences for diagnosing internal and primary fistula tract clarity demonstrated a significant superiority of the multi-phase CE-FS-T1-3D-VIBE sequence in delineating internal clarity (P=0.013) and primary fistula tract clarity (P<0.001). The multi-phase CE-FS-T1-3D-VIBE sequence demonstrated superior accuracy over the FS-T2WI-DWI sequence in depicting localization of internal openings [86.31% vs. 77.38%, 95% confidence interval (CI): 0.307-0.959, P=0.034], as well as delineation of the secondary tracts (88.69% vs. 80.95%, 95% CI: 0.293-1.001, P=0.048). Despite the increased accuracy of the multi-phase CE-FS-T1-3D-VIBE sequence, no difference was observed between the two imaging techniques regarding the accuracy in determining the number of internal openings, the quantity of primary fistulas, and the classification of anal fistulas.

CONCLUSIONS: This study elucidates that the multi-phase CE-FS-T1-3D-VIBE imaging sequence potentially represents a more effective noninvasive alternative for the precise evaluation of the positioning and clarity of the internal opening, as well as the delineation of primary and secondary fistula tracts in anal fistula patients, compared to the FS-T2WI-DWI sequence. This enhanced diagnostic capability underscores the utility of multi-phase CE-FS-T1-3D-VIBE in improving the clinical management of anal fistulas.

PMID:39698714 | PMC:PMC11652061 | DOI:10.21037/qims-24-490

Quant Imaging Med Surg. 2024 Dec 5;14(12):8974-8987. doi: 10.21037/qims-24-1097. Epub 2024 Nov 29.

ABSTRACT

BACKGROUND: Preliminary small-sample studies suggest that silent magnetic resonance angiography (MRA) has an advantage over time-of-flight MRA (TOF MRA) in the characterization of brain arteriovenous malformation (BAVM), but did not examine whether the imaging performance of silent MRA was affected by the intrinsic features of BAVM or common clinical factors. This study sought to compare silent MRA and TOF MRA in terms of the visualization and grading of BAVMs in various clinical settings.

METHODS: In total, 85 participants (50 males, 35 females; mean age: 33.5±15.2 years) with BAVM who underwent both silent MRA and TOF MRA using a 3 Tesla (3T) magnetic resonance imaging (MRI) system were consecutively recruited from the Capital Medical University Xuanwu Hospital between April 2020 and October 2022 to participate in this cross-sectional retrospective study. The patients were divided into subgroups according to new hemorrhage presentation, embolization, size, and nidus compactness. Image quality scoring on a 4-point scale, and the accuracy of characteristic visulization and Spetzler-Martin grading were compared between the two MRA techniques and each MRA subgroup using the rank-sum Wilcoxon test and Fisher’s exact test with digital subtraction angiography (DSA) as the reference standard. A multivariable chi-square test was used to examine the interactions between the grouping factors. A P value <0.05 was considered statistically significant.

RESULTS: The average image quality scores were significantly higher for silent MRA than those for TOF MRA overall (2.83±0.42 versus 2.46±0.66, P<0.001) and in each subgroup (P<0.05). For silent MRA, the average image quality score for BAVM in each subgroup did not differ significantly (P>0.05). For TOF MRA, the image quality scores for the new hemorrhage, small nidus, and diffuse nidus groups was significantly reduced (P=0.001, <0.001, and 0.037, respectively). The accuracy of silent MRA was significantly better than that of TOF MRA in terms of nidus size and Spetzler-Martin grading (P<0.001), but did not differ significantly in terms of deep venous drainage and associated aneurysm (P=0.402, 0.098, respectively). In relation to silent MRA, the image quality, detection of BAVM characteristics, and grading were similar across the new hemorrhage, embolization, size, and compactness subgroups (P=0.066-0.959). In relation to TOF MRA, the accuracy of nidus size grading was significantly lower in the medium-size subgroup than the small-size subgroup (P<0.001).

CONCLUSIONS: Silent MRA performed well in imaging BAVM, and high performance in determining nidus size and Spetzler-Martin grading, but its ability to detect deep venous drainage was limited.

PMID:39698707 | PMC:PMC11651979 | DOI:10.21037/qims-24-1097

Quant Imaging Med Surg. 2024 Dec 5;14(12):8183-8195. doi: 10.21037/qims-24-278. Epub 2024 Oct 21.

ABSTRACT

BACKGROUND: Early recognition of sinus invasion for meningiomas matters the clinical intervention. Therefore, we retrospectively investigated the relationship between peritumoral vessel features on magnetic resonance venography (MRV) and sinus invasion status.

METHODS: Images of phase contrast MRV (PC-MRV, n=46) and contrast-enhanced MRV (CE-MRV, n=39) were independently assessed by four experienced neuroradiologists, including the adjacent sinus status, the peritumoral vessel count and diameter-associated parameters. The sinus invasion status confirmed based on Sindou’s criteria during surgery was taken as the gold standard. The relationship between these MRV-based vessel characteristics and sinus invasion status was further analyzed.

RESULTS: The judgment of sinus invasion based on PC-MRV (n=46) and CE-MRV (n=39) revealed a total accuracy of 63% and 74.4%, respectively. The MRV-based vessel count and associated diameter parameters demonstrated statistical differences between the non-invasion and the invasion group (P<0.05). Under the cutoff value of 3 vessels on PC-MRV and 5.5 vessels on CE-MRV, the prediction of sinus invasion status finally achieved the accuracy of 69.6% and 84.6%, respectively. Furthermore, the vessel count, the sum vascular diameter and the max vascular diameter remained significantly different in further subgroup analyses. A comprehensive generalized linear models (GLM) model based on MRV-related vascular features showed the best diagnostic performance on sinus invasion, with the area under the receiver operating characteristic curve of 0.859 for PC-MRV and 0.913 for CE-MRV.

CONCLUSIONS: For para-sinus meningioma, peritumoral vessel characteristics on MRV, especially the vessel count, exhibited close relationships with sinus status, and thus could be a novel tool to predict sinus invasion.

PMID:39698705 | PMC:PMC11652032 | DOI:10.21037/qims-24-278

Quant Imaging Med Surg. 2024 Dec 5;14(12):8479-8488. doi: 10.21037/qims-24-43. Epub 2024 Nov 29.

ABSTRACT

BACKGROUND: Video-assisted thoracoscopic surgery (VATS) has been widely used for the resection of lung lesions. However, it is difficult to palpate or see small lesions, especially ground-glass opacities (GGOs) during VATS. Thoracoscopic ultrasound has definite value in locating pulmonary parenchymal nodules. However, due to the air in the lung parenchyma, its wide application is limited. This study investigated the value of thoracoscopic ultrasound for the localization of GGOs with incomplete lung collapse in VATS.

METHODS: A retrospective analysis was conducted on patients diagnosed with ground-glass nodules (GGNs) on computed tomography (CT) at Liaoning Province Tumor Hospital from November 2018 to August 2019, who underwent thoracoscopic ultrasound localization and VATS. Screening was conducted for patients who did not achieve complete collapse of the lungs after natural collapse during surgery, the success rate was calculated, and preoperative CT features and thoracoscopic ultrasound features of GGNs were summarized and analyzed.

RESULTS: The success rate of GGOs’ localization by thoracoscopic ultrasound in incomplete collapse lung was 56.67%. Of all preoperative CT features, only the distance between the nodule and the pleura was statistically different (P=0.001). The closer the GGO was to the pleura, the easier it was to be detected.

CONCLUSIONS: Thoracoscopic ultrasound could effectively locate GGO in the condition of incomplete lung collapse and not affect the surgical field, especially when the GGO is close to the pleura, which provides an alternative method for GGO localization in patients who cannot achieve complete lung collapse during VATS.

PMID:39698688 | PMC:PMC11652022 | DOI:10.21037/qims-24-43

Quant Imaging Med Surg. 2024 Dec 5;14(12):8927-8941. doi: 10.21037/qims-24-907. Epub 2024 Nov 29.

ABSTRACT

BACKGROUND: Deep learning has developed rapidly, and deep learning reconstruction (DLR) methods in magnetic resonance imaging (MRI) are gaining attention for their potential to improve efficacy in clinical work. The preoperative MRI assessment of rectal cancer is crucial for patient management, but the imaging quality is currently limited by a number of factors. DLR could be applied to the preoperative MRI assessment of primary rectal cancer, but research about its specific reliability is limited. Thus, this study aimed to evaluate the reliability of DLR in the preoperative MRI examination of primary rectal cancer.

METHODS: This cross-sectional study was conducted at Ruijin Hospital, Shanghai Jiaotong University School of Medicine from March 2022 to October 2022. Patients with primary rectal cancer underwent routine MRI scans on a 3.0T magnetic resonance scanner (SIGNA Architect, GE Healthcare, USA) with 32-channels flexible coil with conventional reconstruction (ConR) and DLR. The DLR method had three noise reduction levels: DLR-H: 75% noise reduction reconstruction; DLR-M: 50% noise reduction reconstruction; and DLR-L: 25% noise reduction reconstruction. Three components were evaluated: objective image quality; subjective image quality; and diagnostic performance. The objective image quality assessment included the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). The subjective image quality assessment involved evaluating five subjective image quality parameters based on a 4-point Likert scale. The diagnostic performance assessment included tumour (T) staging, node (N) staging, as well as the circumferential resection margin and extramural vascular invasion evaluation. The images were evaluated in a blinded manner by two radiologists with different levels of experience. The paired sample Wilcoxon signed-rank test, Kappa test, interclass correlation coefficient, Chi-square test, Friedman test, and weighted kappa coefficients were used for the statistical analysis.

RESULTS: In total, 61 patients (mean age: 65±12 years; 38 men) were enrolled in the study. The DLR method improved the SNR and CNR values of the images relative to the ConR method, while the DLR-H produced the greatest improvement (P<0.040). The subjective image quality of the DLR-H images was superior to that of the ConR images (P<0.001), but there was no significant difference between the DLR-H and DLR-M images (P≥0.075). The evaluators showed good agreement in subjective scoring, and in the DLR image scoring, the evaluators have the best consistency in the DLR-H images scoring (kappa =0.921, P<0.001). The diagnostic efficacy of the DLR images was comparable to that of the ConR images in terms of T staging [Reader 1 (R1): P=0.603; Reader 2 (R2): P=0.206] and N staging (R1: P=0.990; R2: P=0.884).

CONCLUSIONS: The DLR method improved the quality of the images, and had comparable diagnostic efficacy without additional scanning time to that of the ConR method, and thus could be a feasible option for replacing the ConR method in the preoperative MRI examination of primary rectal cancer.

PMID:39698686 | PMC:PMC11651964 | DOI:10.21037/qims-24-907

Quant Imaging Med Surg. 2024 Dec 5;14(12):9506-9521. doi: 10.21037/qims-24-1138. Epub 2024 Nov 29.

ABSTRACT

BACKGROUND: Current medical image registration methods based on Transformer still encounter challenges, including significant local intensity differences and limited computational efficiency when dealing with three-dimensional (3D) computed tomography (CT) and cone beam CT (CBCT) images. These limitations hinder the precise alignment necessary for effective diagnosis and treatment planning. Therefore, the aim of this study is to develop a novel method that overcomes these challenges by enhancing feature interaction and computational efficiency in 3D medical image registration.

METHODS: This paper introduces a novel method that enhances feature interaction within Transformer by computing attention within resizable spatial perpendicular window (SPW). Additionally, it introduces a self-learning mapping control (SLMC) mechanism, which uses a mini convolutional neural network (CNN) to adaptively transform feature vectors into probability vectors. This approach is integrated into the UNet framework, resulting in the SPW-TransUNet. The effectiveness of the SPW-TransUNet is demonstrated through evaluations on two critical 3D medical imaging tasks: CT-CBCT registration and inter-CT registration. We utilized a range of evaluation metrics including Dice similarity coefficient (DICE), structural similarity index measure (SSIM), target registration error (TRE), and negative Jacobian percentage. The validation process involved comparative analysis against established baseline methods using statistical tests to ensure the robustness and reliability of our results.

RESULTS: The proposed method demonstrated outstanding performance in the registration of 124 pairs of CT-CBCT lung images from 20 patients, achieving the lowest TRE of 2.16 mm and a minimal negative Jacobian of 0.126. It also recorded the highest SSIM and Dice coefficient of 86.87% and 88.28%, respectively. For the liver CT task involving 150 patients, the method achieved peak SSIM and DICE scores of 76.92% and 85.77%, respectively. Furthermore, ablation studies confirmed the effectiveness of the designed structural components.

CONCLUSIONS: The SPW-TransUNet offers significant improvements in feature interaction and computational efficiency for medical image registration, providing an effective reference solution for patient and target localization in image-guided radiation therapy.

PMID:39698684 | PMC:PMC11651926 | DOI:10.21037/qims-24-1138