Automatic Skeleton Segmentation in CT Images Based on U-Net.

Bone metastasis, emerging oncological therapies, and osteoporosis represent some of the distinct clinical contexts which can result in morphological alterations in bone structure. The visual assessment of these changes through anatomical images is considered suboptimal, emphasizing the importance of precise skeletal segmentation as a valuable aid for its evaluation. In the present study, a neural network model for automatic skeleton segmentation from bidimensional computerized tomography (CT) slices is proposed. A total of 77 CT images and their semimanual skeleton segmentation from two acquisition protocols (whole-body and femur-to-head) are used to form a training group and a testing group. Preprocessing of the images includes four main steps: stretcher removal, thresholding, image clipping, and normalization (with two different techniques: interpatient and intrapatient). Subsequently, five different sets are created and arranged in a randomized order for the training phase. A neural network model based on U-Net architecture is implemented with different values of the number of channels in each feature map and number of epochs. The model with the best performance obtains a Jaccard index (IoU) of 0.959 and a Dice index of 0.979. The resultant model demonstrates the potential of deep learning applied in medical images and proving its utility in bone segmentation.

Robustness of [18F]FDG PET/CT radiomic analysis in the setting of drug-induced cardiotoxicity.

BACKGROUND AND OBJECTIVES: Standardization of radiomic data acquisition protocols is still at a very early stage, revealing a strong need to work towards the definition of uniform image processing methodologies The aim of this study is to identify sources of variability in radiomic data derived from image discretization and resampling methodologies prior to image feature extraction. Furthermore, to identify robust potential image-based biomarkers for the early detection of cardiotoxicity. METHODS: Image post-acquisition processing, interpolation, and volume of interest (VOI) segmentation were performed. Four experiments were conducted to assess the reliability in terms of the intraclass correlation coefficient (ICC) of the radiomic features and the effects of the variation of voxel size and gray level discretization. Statistical analysis was performed separating the patients according to cardiotoxicity diagnosis. Differences of texture features were studied with Mann-Whitney U test. P-values <0.05 after multiple testing correction were considered statistically significant. Additionally, a non-supervised k-Means clustering algorithm was evaluated. RESULTS: The effect of the variation in the voxel size demonstrated a non-dependency relationship with the values of the radiomic features, regardless of the chosen discretization method. The median ICC values were 0.306 and 0.872 for absolute agreement and consistency, respectively, when varying the discretization bin number. The median ICC values were 0.678 and 0.878 for absolute agreement and consistency, respectively, when varying the discretization bin size. A total of 16 first order, 6 Gray Level Co-occurrence Matrix (GLCM), 4 Gray Level Dependence Matrix (GLDM) and 4 Gray Level Run Length Matrix (GLRLM) features demonstrated statistically significant differences between the diagnosis groups for interim scans (P<0.05) for the fixed bin size (FBS) discretization methodology. However, no statistically significant differences between diagnostic groups were found for the fixed bin number (FBN) discretization methodology. Two clusters based on the radiomic features were identified. CONCLUSIONS: Gray level discretization has a major impact on the repeatability of the radiomic features. The selection of the optimal processing methodology has led to the identification of texture-based patterns for the differentiation of early cardiac damage profiles.

Radiomics analysis of bone marrow biopsy locations in [18F]FDG PET/CT images for measurable residual disease assessment in multiple myeloma.

The combination of visual assessment of whole body [18F]FDG PET images and evaluation of bone marrow samples by Multiparameter Flow Cytometry (MFC) or Next-Generation Sequencing (NGS) is currently the most common clinical practice for the detection of Measurable Residual Disease (MRD) in Multiple Myeloma (MM) patients. In this study, radiomic features extracted from the bone marrow biopsy locations are analyzed and compared to those extracted from the whole bone marrow in order to study the representativeness of these biopsy locations in the image-based MRD assessment. Whole body [18F]FDG PET of 39 patients with newly diagnosed MM were included in the database, and visually evaluated by experts in nuclear medicine. A methodology for the segmentation of biopsy sites from PET images, including sternum and posterior iliac crest, and their subsequent quantification is proposed. First, starting from the bone marrow segmentation, a segmentation of the biopsy sites is performed. Then, segmentations are quantified extracting SUV metrics and radiomic features from the [18F]FDG PET images and are evaluated by Mann-Whitney U-tests as valuable features differentiating PET+/PET- and MFC+ /MFC- groups. Moreover, correlation between whole bone marrow and biopsy sites is studied by Spearman ρ rank. Classification performance of the radiomics features is evaluated applying seven machine learning algorithms. Statistical analyses reveal that some images features are significant in PET+/PET- differentiation, such as SUVmax, Gray Level Non-Uniformity or Entropy, especially with a balanced database where 16 of the features show a p value < 0.001. Correlation analyses between whole bone marrow and biopsy sites results in significant and acceptable coefficients, with 11 of the variables reaching a correlation coefficient greater than 0.7, with a maximum of 0.853. Machine learning algorithms demonstrate high performances in PET+/PET- classification reaching a maximum AUC of 0.974, but not for MFC+/MFC- classification. The results demonstrate the representativeness of sample sites as well as the effectiveness of extracted features (SUV metrics and radiomic features) from the [18F]FDG PET images in MRD assessment in MM patients.

Static first-minute-frame (FMF) PET imaging after 18F-labeled amyloid tracer injection is correlated to [18F]FDG PET in patients with primary progressive aphasia.

OBJECTIVE: To study the correlation between a static PET image of the first-minute-frame (FMF) acquired with 18F-labeled amyloid-binding radiotracers and brain [18F]FDG PET in patients with primary progressive aphasia (PPA). MATERIAL AND METHODS: The study cohort includes 17 patients diagnosed with PPA with the following distribution: 9 nonfluent variant PPA, 4 logopenic variant PPA, 1 semantic variant PPA, 3 unclassifiable PPA. Regional SUVRs are extracted from FMFs and their corresponding [18F]FDG PET images and Pearson's correlation coefficients are calculated. RESULTS: SUVRs of both images show similar patterns of regional cerebral alterations. Intrapatient correlation analyses result in a mean coefficient of r=0.94±0.06. Regional interpatient correlation coefficients of the study cohort are greater than 0.81. Radiotracer-specific and variant-specific subcohorts show no difference in the similarity between the images. CONCLUSIONS: The static FMF could be a valid alternative to dynamic early-phase amyloid PET proposed in the literature, and a neurodegeneration biomarker for the diagnosis and classification of PPA in amyloid PET studies.

CASSIA (cardiology software suite for image analysis): a potential new tool for the evaluation of [18F]FDG PET/CT in the setting of infective endocarditis.

PURPOSE: Due to the high morbidity and mortality of infective endocarditis (IE), medical imaging techniques are combined to ensure a correct diagnosis. [18F]FDG PET/CT has demonstrated the ability to improve diagnostic accuracy compared with the conventional modified Duke criteria in patients with suspected IE, especially those with prosthetic valve infective endocarditis (PVIE). The aim of this study is to provide an adjunctive diagnostic tool to improve the diagnostic accuracy in cardiovascular infections, specifically PVIE. METHODS: A segmentation tool to extract quantitative measures of [18F]FDG PET/CT image studies of prosthetic heart valve regions was developed and validated in 20 cases of suspected PVIE, of which 9 were confirmed. For that, Valvular Heterogeneity Index (VHI) and Ring-to-Center Ratio (RCR) were defined. RESULTS: Results show an overall increase in the metabolic uptake of the prosthetic valve ring in the studies with confirmed PVIE diagnosis (SUVmax from 1.70 to 3.20; SUVmean from 0.86 to 1.50). The VHI and RCR showed areas under the curve of 0.727 and 0.808 in the receiver operating characteristics curve analyses, respectively, for PVIE diagnosis. Mann-Whitney U tests showed statistically significant differences between groups for RCR (p = 0.02). Visual analyses and clinical reports were concordant with the extracted quantitative metrics. CONCLUSION: The proposed new method and presented software solution (CASSIA) provide the capability to assess quantitatively myocardial metabolism along the prosthetic valve region in routine [18F]FDG PET/CT scans for evaluating heart valve infectious processes. VHI and RCR are proposed as new potential adjunctive measures for PVIE diagnosis.

Bone marrow segmentation and radiomics analysis of [18F]FDG PET/CT images for measurable residual disease assessment in multiple myeloma.

BACKGROUND AND OBJECTIVES: The last few years have been crucial in defining the most appropriate way to quantitatively assess [18F]FDG PET images in Multiple Myeloma (MM) patients to detect persistent tumor burden. The visual evaluation of images complements the assessment of Measurable Residual Disease (MRD) in bone marrow samples by multiparameter flow cytometry (MFC) or next-generation sequencing (NGS). The aim of this study was to quantify MRD by analyzing quantitative and texture [18F]FDG PET features. METHODS: Whole body [18F]FDG PET of 39 patients with newly diagnosed MM were included in the database, and visually evaluated by experts in nuclear medicine. A segmentation methodology of the skeleton from CT images and an additional manual segmentation tool were proposed, implemented in a software solution including a graphical user interface. Both the compact bone and the spinal canal were removed from the segmentation to obtain only the bone marrow mask. SUV metrics, GLCM, GLRLM, and NGTDM parameters were extracted from the PET images and evaluated by Mann-Whitney U-tests and Spearman ρ rank correlation as valuable features differentiating PET+/PET- and MFC+/MFC- groups. Seven machine learning algorithms were applied for evaluating the classification performance of the extracted features. RESULTS: Quantitative analysis for PET+/PET- differentiating demonstrated to be significant for most of the variables assessed with Mann-Whitney U-test such as Variance, Energy, and Entropy (p-value = 0.001). Moreover, the quantitative analysis with a balanced database evaluated by Mann-Whitney U-test revealed in even better results with 19 features with p-values < 0.001. On the other hand, radiomics analysis for MFC+/MFC- differentiating demonstrated the necessity of combining MFC evaluation with [18F]FDG PET assessment in the MRD diagnosis. Machine learning algorithms using the image features for the PET+/PET- classification demonstrated high performance metrics but decreasing for the MFC+/MFC- classification. CONCLUSIONS: A proof-of-concept for the extraction and evaluation of bone marrow radiomics features of [18F]FDG PET images was proposed and implemented. The validation showed the possible use of these features for the image-based assessment of MRD.

The value of metabolic parameters and textural analysis in predicting prognosis in locally advanced cervical cancer treated with chemoradiotherapy.

OBJECTIVE: The aim of the study was to assess the impact of clinical and metabolic parameters derived from 18F-FDG PET/CT (positron emission tomography-computed tomography) in patients with locally advanced cervical cancer (LACC) on prognosis. METHODS: Patients with LACC of stage IB2-IVA treated by primary radiochemotherapy followed by brachytherapy were enrolled in this retrospective study. Indexes derived from standardized uptake value (SUV), metabolic tumor volume (MTV), total lesion glycolysis (TLG), and textural features of the primary tumor were measured for each patient. Overall survival (OS) and recurrence-free survival (RFS) rates were calculated according to Kaplan-Meier and survival curves were compared using the log-rank test. Uni- and multivariate analyses were performed using the Cox regression model. RESULTS: A total of 116 patients were included. Median follow-up was 58 months (range: 1-129). A total of 36 (31%) patients died. Five-year OS and RFS rates were 69 and 60%, respectively. Univariate analyses indicated that FIGO stage, the presence of hydronephrosis, high CYFRA 21.1 levels, and textural features had a significant impact on OS and RFS. MTV as well as SCC-Ag concentration were also significantly associated with OS. On multivariate analysis, the presence of hydronephrosis, CYFRA 21.1, and sphericity were independent prognostics factors for OS and RFS. Also, SCC-Ag level, MTV, and GLZLM (gray-level zone length matrix) ZLNU (zone length non-uniformity) were significantly associated with OS. CONCLUSION: Classical prognostic factors and tumor heterogeneity on pretreatment PET/CT were significantly associated with prognosis in patients with LACC.

An image processing tool for the detection of anthracycline-induced cardiotoxicity by evaluating the myocardial metabolic activity in [18F]FDG PET/CT.

PURPOSE: Chemotherapy-induced cardiotoxicity is one of the main complications during and after cancer treatment. While echocardiography is the most used technique in clinical practice to evaluate left ventricular (LV) dysfunction, a multimodal approach is preferred for the early detection of anthracycline-induced cardiotoxicity. In this paper, an image processing tool allowing the qualitative and quantitative analysis of myocardial metabolic activity by [18F]fluorodeoxyglucose (FDG) positron emission tomography computed tomography (PET/CT) images, acquired routinely during and after cancer treatment, is presented. METHODS: The methodology is based on cardiac single photon emission computed tomography image processing protocols used in clinical practice. LV polar maps are created, and quantitative regional values are calculated. The tool was validated in a study group of 24 patients with Hodgkin or non-Hodgkin lymphoma (HL and NHL, respectively) treated with anthracyclines. Staging, interim and end-of-treatment [18F]FDG PET/CT images were acquired and the presented tool was used to extract the quantitative metrics of LV metabolic activity. RESULTS: Results show an overall increase of metabolic activity in the interim PET image acquired while on treatment compared to staging PET, which then decreased in the end-of-treatment scan. Positive correlation coefficients between staging and interim scans, and negative correlation coefficients between interim and end-of-treatment scans also support this finding. Metabolic changes occur predominantly in the septal region. CONCLUSION: The proposed methodology and presented software solution provides the capability to assess quantitatively myocardial metabolism acquired by routine [18F]FDG PET/CT scanning during cancer treatment for evaluating anthracycline-induced cardiotoxicity. The [18F]FDG PET/CT septal-lateral uptake ratio is proposed as a new quantitative measure of myocardial metabolism.

Differences in Striatal Metabolism in [18F]FDG PET in Parkinson's Disease and Atypical Parkinsonism.

Neurodegenerative parkinsonisms affect mainly cognitive and motor functions and are syndromes of overlapping symptoms and clinical manifestations such as tremor, rigidness, and bradykinesia. These include idiopathic Parkinson's disease (PD) and the atypical parkinsonisms, namely progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), multiple system atrophy (MSA) and dementia with Lewy body (DLB). Differences in the striatal metabolism among these syndromes are evaluated using [18F]FDG PET, caused by alterations to the dopaminergic activity and neuronal loss. A study cohort of three patients with PD, 29 with atypical parkinsonism (10 PSP, 6 CBD, 2 MSA, 7 DLB, and 4 non-classifiable), and a control group of 25 patients with normal striatal metabolism is available. Standardized uptake value ratios (SUVR) are extracted from the striatum, and the caudate and the putamen separately. SUVRs are compared among the study groups. In addition, hemispherical and caudate-putamen differences are evaluated in atypical parkinsonisms. Striatal hypermetabolism is detected in patients with PD, while atypical parkinsonisms show hypometabolism, compared to the control group. Hemispherical differences are observed in CBD, MSA and DLB, with the latter also showing statistically significant caudate-putamen asymmetry (p = 0.018). These results indicate disease-specific metabolic uptake patterns in the striatum that can support the differential diagnosis.

High Correlation of Static First-Minute-Frame (FMF) PET Imaging after 18F-Labeled Amyloid Tracer Injection with [18F]FDG PET Imaging.

Dynamic early-phase PET images acquired with radiotracers binding to fibrillar amyloid-beta (Aß) have shown to correlate with [18F]fluorodeoxyglucose (FDG) PET images and provide perfusion-like information. Perfusion information of static PET scans acquired during the first minute after radiotracer injection (FMF, first-minute-frame) is compared to [18F]FDG PET images. FMFs of 60 patients acquired with [18F]florbetapir (FBP), [18F]flutemetamol (FMM), and [18F]florbetaben (FBB) are compared to [18F]FDG PET images. Regional standardized uptake value ratios (SUVR) are directly compared and intrapatient Pearson's correlation coefficients are calculated to evaluate the correlation of FMFs to their corresponding [18F]FDG PET images. Additionally, regional interpatient correlations are calculated. The intensity profiles of mean SUVRs among the study cohort (r = 0.98, p < 0.001) and intrapatient analyses show strong correlations between FMFs and [18F]FDG PET images (r = 0.93 ± 0.05). Regional VOI-based analyses also result in high correlation coefficients. The FMF shows similar information to the cerebral metabolic patterns obtained by [18F]FDG PET imaging. Therefore, it could be an alternative to the dynamic imaging of early phase amyloid PET and be used as an additional neurodegeneration biomarker in amyloid PET studies in routine clinical practice while being acquired at the same time as amyloid PET images.

Deep Learning Automated Segmentation for Muscle and Adipose Tissue from Abdominal Computed Tomography in Polytrauma Patients.

Manual segmentation of muscle and adipose compartments from computed tomography (CT) axial images is a potential bottleneck in early rapid detection and quantification of sarcopenia. A prototype deep learning neural network was trained on a multi-center collection of 3413 abdominal cancer surgery subjects to automatically segment truncal muscle, subcutaneous adipose tissue and visceral adipose tissue at the L3 lumbar vertebral level. Segmentations were externally tested on 233 polytrauma subjects. Although after severe trauma abdominal CT scans are quickly and robustly delivered, with often motion or scatter artefacts, incomplete vertebral bodies or arms that influence image quality, the concordance was generally very good for the body composition indices of Skeletal Muscle Radiation Attenuation (SMRA) (Concordance Correlation Coefficient (CCC) = 0.92), Visceral Adipose Tissue index (VATI) (CCC = 0.99) and Subcutaneous Adipose Tissue Index (SATI) (CCC = 0.99). In conclusion, this article showed an automated and accurate segmentation system to segment the cross-sectional muscle and adipose area L3 lumbar spine level on abdominal CT. Future perspectives will include fine-tuning the algorithm and minimizing the outliers.