Segmentation of nasopharyngeal carcinoma (NPC) lesions in MR images

DCTR U-Net: automatic segmentation algorithm for medical images of nasopharyngeal cancer in the context of deep learning

Nasopharyngeal carcinoma (NPC) is a malignant tumor that occurs in the wall of the nasopharyngeal cavity and is prevalent in Southern China, Southeast Asia, North Africa, and the Middle East. According to studies, NPC is one of the most common malignant tumors in Hainan, China, and it has the highest incidence rate among otorhinolaryngological malignancies. We proposed a new deep learning network model to improve the segmentation accuracy of the target region of nasopharyngeal cancer. Our model is based on the U-Net-based network, to which we add Dilated Convolution Module, Transformer Module, and Residual Module. The new deep learning network model can effectively solve the problem of restricted convolutional fields of perception and achieve global and local multi-scale feature fusion. In our experiments, the proposed network was trained and validated using 10-fold cross-validation based on the records of 300 clinical patients. The results of our network were evaluated using the dice similarity coefficient (DSC) and the average symmetric surface distance (ASSD). The DSC and ASSD values are 0.852 and 0.544 mm, respectively. With the effective combination of the Dilated Convolution Module, Transformer Module, and Residual Module, we significantly improved the segmentation performance of the target region of the NPC.

Predicting prognosis of nasopharyngeal carcinoma based on deep learning: peritumoral region should be valued

BACKGROUND

The purpose of this study was to explore whether incorporating the peritumoral region to train deep neural networks could improve the performance of the models for predicting the prognosis of NPC.

METHODS

A total of 381 NPC patients who were divided into high- and low-risk groups according to progression-free survival were retrospectively included. Deeplab v3 and U-Net were trained to build segmentation models for the automatic segmentation of the tumor and suspicious lymph nodes. Five datasets were constructed by expanding 5, 10, 20, 40, and 60 pixels outward from the edge of the automatically segmented region. Inception-Resnet-V2, ECA-ResNet50t, EfficientNet-B3, and EfficientNet-B0 were trained with the original, segmented, and the five new constructed datasets to establish the classification models. The receiver operating characteristic curve was used to evaluate the performance of each model.

RESULTS

The Dice coefficients of Deeplab v3 and U-Net were 0.741(95%CI:0.722-0.760) and 0.737(95%CI:0.720-0.754), respectively. The average areas under the curve (aAUCs) of deep learning models for classification trained with the original and segmented images and with images expanded by 5, 10, 20, 40, and 60 pixels were 0.717 ± 0.043, 0.739 ± 0.016, 0.760 ± 0.010, 0.768 ± 0.018, 0.802 ± 0.013, 0.782 ± 0.039, and 0.753 ± 0.014, respectively. The models trained with the images expanded by 20 pixels obtained the best performance.

CONCLUSIONS

The peritumoral region NPC contains information related to prognosis, and the incorporation of this region could improve the performance of deep learning models for prognosis prediction.

Computer-aided diagnosis and regional segmentation of nasopharyngeal carcinoma based on multi-modality medical images

PURPOSE

Nasopharyngeal carcinoma (NPC) is a category of tumors with high incidence in head-and-neck (H&N) body region, and the diagnosis and treatment planning are usually conducted by radiologists manually, which is tedious, time-consuming and unrepeatable. In this paper, we integrated different stages of this process and proposed a computer-aided framework to realize automatic detection, tumor region and sub-region segmentation, and visualization of NPC, which are usually investigated separately in literatures.

METHODS

Multi-modality images are utilized in the framework. Firstly, NPC is detected by a convolutional neural network (CNN) on computed tomography (CT) scans. Then, NPC area is segmented from magnetic resonance imaging (MRI) images by using a multi-modality MRI fusion network. Thirdly, NPC sub-regions with different metabolic activities are divided on CT images of the same patient via an adaptive threshold algorithm. Finally, 3D surface model of NPC is generated for observing its shape, size, and location in the head region. The proposed method is compared with other algorithms by evaluation on the volumes and shapes of detected NPCs.

RESULTS

Experiments are conducted on CT images of 130 NPC patients and 102 subjects without NPC and MRI images of 149 NPC patients, among which 52 subjects are overlapped with both CT and MRI images. The reference for evaluation is generated by three experienced radiologists. The results demonstrated that our utilized models outperform other strategies with detection accuracy 0.882 and Dice similarity coefficient 0.719 for NPC segmentation. Sub-region division and the 3D visualized models show great acceptability in clinical usage.

CONCLUSION

The remarkable performance indicated the potential of our framework in alleviating workload of radiologist. Furthermore, the combined usage of multi-modality images is able to generate reliable segmentations of NPC area and sub-regions, which provide evidence to judge the heterogeneity among patients and guide the dose painting for radiation therapy.

A Collaborative Dictionary Learning Model for Nasopharyngeal Carcinoma Segmentation on Multimodalities MR Sequences

Nasopharyngeal carcinoma (NPC) is the most common malignant tumor of the nasopharynx. The delicate nature of the nasopharyngeal structures means that noninvasive magnetic resonance imaging (MRI) is the preferred diagnostic technique for NPC. However, NPC is a typically infiltrative tumor, usually with a small volume, and thus, it remains challenging to discriminate it from tightly connected surrounding tissues. To address this issue, this study proposes a voxel-wise discriminate method for locating and segmenting NPC from normal tissues in MRI sequences. The located NPC is refined to obtain its accurate segmentation results by an original multiviewed collaborative dictionary classification (CODL) model. The proposed CODL reconstructs a latent intact space and equips it with discriminative power for the collective multiview analysis task. Experiments on synthetic data demonstrate that CODL is capable of finding a discriminative space for multiview orthogonal data. We then evaluated the method on real NPC. Experimental results show that CODL could accurately discriminate and localize NPCs of different volumes. This method achieved superior performances in segmenting NPC compared with benchmark methods. Robust segmentation results show that CODL can effectively assist clinicians in locating NPC.

Automated nasopharyngeal carcinoma segmentation in magnetic resonance images by combination of convolutional neural networks and graph cut

Accurate and reliable segmentation of nasopharyngeal carcinoma (NPC) in medical images is an import task for clinical applications, including radiotherapy. However, NPC features large variations in lesion size and shape, as well as inhomogeneous intensities within the tumor and similar intensity to that of nearby tissues, making its segmentation a challenging task. The present study proposes a novel automated NPC segmentation method in magnetic resonance (MR) images by combining a deep convolutional neural network (CNN) model and a 3-dimensional (3D) graph cut-based method in a two-stage manner. First, a multi-view deep CNN-based segmentation method is performed. A voxel-wise initial segmentation is generated by integrating the inferential classification information of three trained single-view CNNs. Instead of directly using the CNN classification results to achieve a final segmentation, the proposed method uses a 3D graph cut-based method to refine the initial segmentation. Specifically, the probability response map obtained using the multi-view CNN method is utilized to calculate the region cost, which represents the likelihood of a voxel being assigned to the tumor or non-tumor. Structure information in 3D from the original MR images is used to calculate the boundary cost, which measures the difference between the two voxels in the 3D neighborhood. The proposed method was evaluated on T1-weighted images from 30 NPC patients using the leave-one-out method. The experimental results demonstrated that the proposed method is effective and accurate for NPC segmentation.

Validation of a Magnetic Resonance Imaging-based Auto-contouring Software Tool for Gross Tumour Delineation in Head and Neck Cancer Radiotherapy Planning

AIMS

To carry out statistical validation of a newly developed magnetic resonance imaging (MRI) auto-contouring software tool for gross tumour volume (GTV) delineation in head and neck tumours to assist in radiotherapy planning.

MATERIALS AND METHODS

Axial MRI baseline scans were obtained for 10 oropharyngeal and laryngeal cancer patients. GTV was present on 102 axial slices and auto-contoured using the modified fuzzy c-means clustering integrated with the level set method (FCLSM). Peer-reviewed (C-gold) manual contours were used as the reference standard to validate auto-contoured GTVs (C-auto) and mean manual contours (C-manual) from two expert clinicians (C1 and C2). Multiple geometric metrics, including the Dice similarity coefficient (DSC), were used for quantitative validation. A DSC≥0.7 was deemed acceptable. Inter- and intra-variabilities among the manual contours were also validated. The two-dimensional contours were then reconstructed in three dimensions for GTV volume calculation, comparison and three-dimensional visualisation.

RESULTS

The mean DSC between C-gold and C-auto was 0.79. The mean DSC between C-gold and C-manual was 0.79 and that between C1 and C2 was 0.80. The average time for GTV auto-contouring per patient was 8 min (range 6-13 min; mean 45 s per axial slice) compared with 15 min (range 6-23 min; mean 88 s per axial slice) for C1. The average volume concordance between C-gold and C-auto volumes was 86.51% compared with 74.16% between C-gold and C-manual. The average volume concordance between C1 and C2 volumes was 86.82%.

CONCLUSIONS

This newly designed MRI-based auto-contouring software tool shows initial acceptable results in GTV delineation of oropharyngeal and laryngeal tumours using FCLSM. This auto-contouring software tool may help reduce inter- and intra-variability and can assist clinical oncologists with time-consuming, complex radiotherapy planning.

Unidimensional Measurement May Evaluate Target Lymph Nodal Response After Induction Chemotherapy for Nasopharyngeal Carcinoma

The aim of the study was to evaluate whether short axis and long axis on axial and coronal magnetic resonance imaging planes would reflect the tumor burden or alteration in size after induction chemotherapy in nasopharyngeal carcinoma. Patients with pathologically confirmed nasopharyngeal carcinoma (n = 37) with at least 1 positive cervical lymph node (axial short axis ≥15 mm) were consecutively enrolled in this prospective study. Lymph nodal measurements were performed along its short axis and long axis in both axial and coronal magnetic resonance imaging planes at diagnosis and after 2 cycles of induction chemotherapy. In addition, lymph nodal volumes were automatically calculated in 3D treatment-planning system, which were used as reference standard. Student's t test or nonparametric Mann-Whitney U test was used to compare the continuous quantitative variables. Meanwhile, the κ statistic and McNemar's test were used to evaluate the degree of agreement and discordance in response categorization among different measurements. Axial short axis was significantly associated with volumes at diagnosis (P < 0.001). A good agreement (κ=0.583) was found between axial short axis and volumetric criteria. However, the inconsistent lymph nodal shrinkage in 4 directions was observed. Axial short-axis shrinking was more rapid than the other 3 parameters. Interestingly, when utilizing the alternative planes for unidimensional measurements to assess tumor response, coronal short-axis showed the best concordance (κ=0.792) to the volumes. Axial short axis may effectively reflect tumor burden or change in tumor size in the assessment of target lymph nodal response after induction chemotherapy for nasopharyngeal carcinoma. However, it should be noted that axial short axis may amplify the therapeutic response. In addition, the role of coronal short axis in the assessment of tumor response needs further evaluation.

Involvement of microRNA-24 and DNA methylation in resistance of nasopharyngeal carcinoma to ionizing radiation

Nasopharyngeal carcinoma (NPC) is a malignant tumor originating in the epithelium. Radiotherapy is the standard therapy, but tumor resistance to this treatment reduces the 5-year patient survival rate dramatically. Studies are urgently needed to elucidate the mechanism of NPC radioresistance. Epigenetics--particularly microRNAs (miRNA) and DNA methylation--plays an important role in carcinogenesis and oncotherapy. We used qRT-PCR analysis and identified an miRNA signature from differentially expressed miRNAs. Our objectives were to identify the role of miR24 in NPC tumorigenesis and radioresistance and to identify the mechanisms by which miR24 is regulated. We found that miR24 inhibited NPC cell growth, promoted cell apoptosis, and suppressed the growth of NPC xenografts. We showed that miR24 was significantly downregulated in recurrent NPC tissues. When combined with irradiation, miR24 acted as a radiosensitizer in NPC cells. One of the miR24 precursors was embedded in a CpG island. Aberrant DNA methylation was involved in NPC response to radiotherapy, which linked inactivation of miR24 through hypermethylation of its precursor promoter with NPC radioresistance. Treating NPC cells with the DNA-hypomethylating agent 5-aza-2'-deoxycytidine compensated for the reduced miR24 expression. Together, our findings showed that miR24 was negatively regulated by hypermethylation of its precursor promoter in NPC radioresistance. Our findings defined a central role for miR24 as a tumor-suppressive miRNA in NPC and suggested its use in novel strategies for treatment of this cancer.

Region-based nasopharyngeal carcinoma lesion segmentation from MRI using clustering- and classification-based methods with learning

In clinical diagnosis of nasopharyngeal carcinoma (NPC) lesion, clinicians are often required to delineate boundaries of NPC on a number of tumor-bearing magnetic resonance images, which is a tedious and time-consuming procedure highly depending on expertise and experience of clinicians. Computer-aided tumor segmentation methods (either contour-based or region-based) are necessary to alleviate clinicians' workload. For contour-based methods, a minimal user interaction to draw an initial contour inside or outside the tumor lesion for further curve evolution to match the tumor boundary is preferred, but parameters within most of these methods require manual adjustment, which is technically burdensome for clinicians without specific knowledge. Therefore, segmentation methods with a minimal user interaction as well as automatic parameters adjustment are often favored in clinical practice. In this paper, two region-based methods with parameters learning are introduced for NPC segmentation. Two hundred fifty-three MRI slices containing NPC lesion are utilized for evaluating the performance of the two methods, as well as being compared with other similar region-based tumor segmentation methods. Experimental results demonstrate the superiority of adopting learning in the two introduced methods. Also, they achieve comparable segmentation performance from a statistical point of view.

Quantitative characterization and identification of lymph nodes and nasopharingeal carcinoma by coregistered magnetic resonance images

In this study we developed a technique to improve the identification of carcinoma and pathological lymph nodes in cases of Nasopharingeal Carcinoma (NPC), through a quantitative characterization of the tissues based on MR images: 3D VIBE (Volumetric Interpolated Breath-hold Examination) T1-CE (Contrast Enhanced), T1, T2 and Diffusion Weighted Imaging (DWI) for b-values 0,300,500,700,1000. The procedure included two phases: 1) coregistration of volumes and 2) tissue characterization. Concerning the first phase, the DICOM images were reassembled spatially and resampled with isotropic 0.5mm resolution. Coregistration was performed by two multiresolution rigid transformations, merging head and neck volumes, plus a final multiresolution non rigid transformation. The anatomical 3D CE-VIBE volume was taken as reference. The procedure for tissue characterization is semi automated and it required a radiologist to identify an example of tissue from the primary tumor and a metastatic lymph node. We generated a 8-dimensional membership function to perform a fuzzy-like identification of these tissues. The result of this procedure was the generation of two maps, which showed complementary characterization of lymph nodes and carcinoma. A few example will be shown to evidence the potentiality of this method in identification and characterization of NPC lesions.

Semi-automated segmentation of the sigmoid and descending colon for radiotherapy planning using the fast marching method

A fast and accurate segmentation of organs at risk, such as the healthy colon, would be of benefit for planning of radiotherapy, in particular in an adaptive scenario. For the treatment of pelvic tumours, a great challenge is the segmentation of the most adjacent and sensitive parts of the gastrointestinal tract, the sigmoid and descending colon. We propose a semi-automated method to segment these bowel parts using the fast marching (FM) method. Standard 3D computed tomography (CT) image data obtained from routine radiotherapy planning were used. Our pre-processing steps distinguish the intestine, muscles and air from connective tissue. The core part of our method separates the sigmoid and descending colon from the muscles and other segments of the intestine. This is done by utilizing the ability of the FM method to compute a specified minimal energy functional integrated along a path, and thereby extracting the colon centre line between user-defined control points in the sigmoid and descending colon. Further, we reconstruct the tube-shaped geometry of the sigmoid and descending colon by fitting ellipsoids to points on the path and by adding adjacent voxels that are likely voxels belonging to these bowel parts. Our results were compared to manually outlined sigmoid and descending colon, and evaluated using the Dice coefficient (DC). Tests on 11 patients gave an average DC of 0.83 (+/-0.07) with little user interaction. We conclude that the proposed method makes it possible to fast and accurately segment the sigmoid and descending colon from routine CT image data.

Head and neck cancers on CT: preliminary study of treatment response assessment based on computerized volume analysis

OBJECTIVE

The objective of our study was to investigate the feasibility of computerized segmentation of lesions on head and neck CT scans and evaluate its potential for estimating changes in tumor volume in response to treatment of head and neck cancers.

MATERIALS AND METHODS

Twenty-six CT scans were retrospectively collected from the files of 13 patients with 35 head and neck lesions. The CT scans were obtained from an examination performed before treatment (pretreatment scan) and an examination performed after one cycle of chemotherapy (posttreatment scan). Thirteen lesions were primary site cancers and 22 were metastatic lymph nodes. An experienced radiologist (radiologist 1) marked the 35 lesions and outlined each lesion's 2D contour on the best slice on both the pre- and posttreatment scans. Full 3D contours were also manually extracted for the 13 primary tumors. Another experienced radiologist (radiologist 2) verified and modified, if necessary, all manually drawn 2D and 3D contours. An in-house-developed computerized system performed 3D segmentation based on a level set model.

RESULTS

The computer-estimated change in tumor volume and percentage change in tumor volume between the pre- and posttreatment scans achieved a high correlation (intraclass correlation coefficient [ICC] = 0.98 and 0.98, respectively) with the estimates from manual segmentation for the 13 primary tumors. The average error in estimating the percentage change in tumor volume by automatic segmentation relative to the radiologists' average error was -1.5% +/- 5.4% (SD). For the 35 lesions, the ICC between the automatic and manual estimates of change in pre- to posttreatment tumor area was 0.93 and of percentage change in pre- to posttreatment tumor area was 0.85. The average error in estimating the percentage change in tumor area by automatic segmentation was -3.2% +/- 15.3%.

CONCLUSION

Preliminary results indicate that this computerized segmentation system can reliably estimate changes in tumor size on CT scans relative to radiologists' manual segmentation. This information can be used to calculate changes in tumor size on pre- and posttreatment scans to assess response to treatment.

Automated volume analysis of head and neck lesions on CT scans using 3D level set segmentation

The authors have developed a semiautomatic system for segmentation of a diverse set of lesions in head and neck CT scans. The system takes as input an approximate bounding box, and uses a multistage level set to perform the final segmentation. A data set consisting of 69 lesions marked on 33 scans from 23 patients was used to evaluate the performance of the system. The contours from automatic segmentation were compared to both 2D and 3D gold standard contours manually drawn by three experienced radiologists. Three performance metric measures were used for the comparison. In addition, a radiologist provided quality ratings on a 1 to 10 scale for all of the automatic segmentations. For this pilot study, the authors observed that the differences between the automatic and gold standard contours were larger than the interobserver differences. However, the system performed comparably to the radiologists, achieving an average area intersection ratio of 85.4% compared to an average of 91.2% between two radiologists. The average absolute area error was 21.1% compared to 10.8%, and the average 2D distance was 1.38 mm compared to 0.84 mm between the radiologists. In addition, the quality rating data showed that, despite the very lax assumptions made on the lesion characteristics in designing the system, the automatic contours approximated many of the lesions very well.

Nasopharyngeal Cancers: Which Method Should be Used to Measure these Irregularly Shaped Tumors on Cross-Sectional Imaging?

PURPOSE

To determine whether the standard techniques of measuring tumor size and change in size after treatment could be applied to the measurement of nasopharyngeal cancers, which are often irregular in shape.

METHODS AND MATERIALS

The standard measurements of bidimensional (BDM) (World Health Organization criteria) and unidimensional (UDM) (Response Evaluation Criteria in Solid Tumors [RECIST] criteria), together with the maximum depth of the tumor perpendicular to the pharyngeal wall (DM), were acquired from axial magnetic resonance images of primary nasopharyngeal carcinoma in 44 patients at diagnosis and in 29 of these patients after treatment. Tumor volume measurements (VM), acquired from the summation of areas from the axial magnetic resonance images, were used as the reference standard.

RESULTS

There was a significant association between VM and BDM with respect to tumor size at diagnosis (p = 0.002), absolute change in tumor size after treatment (p < 0.001), and percentage change in tumor size after treatment (p = 0.044), but not between VM and UDM. There was also a significant association between VM and DM with respect to percentage change in tumor size after treatment (p = <0.0001) but not absolute change (p = 0.222).

CONCLUSION

When using simple measurements to assess irregularly shaped nasopharyngeal cancers, the BDM should be used to measure size at diagnosis and the BDM and percentage change in size with treatment. Unidimensional measurement does not reflect size or change in size, and therefore the RECIST criteria may not be applicable to all tumor shapes. The use of DM requires further evaluation.

Automatic Segmentation of Pelvic Structures From Magnetic Resonance Images for Prostate Cancer Radiotherapy

PURPOSE

Target-volume and organ-at-risk delineation is a time-consuming task in radiotherapy planning. The development of automated segmentation tools remains problematic, because of pelvic organ shape variability. We evaluate a three-dimensional (3D), deformable-model approach and a seeded region-growing algorithm for automatic delineation of the prostate and organs-at-risk on magnetic resonance images.

METHODS AND MATERIALS

Manual and automatic delineation were compared in 24 patients using a sagittal T2-weighted (T2-w) turbo spin echo (TSE) sequence and an axial T1-weighted (T1-w) 3D fast-field echo (FFE) or TSE sequence. For automatic prostate delineation, an organ model-based method was used. Prostates without seminal vesicles were delineated as the clinical target volume (CTV). For automatic bladder and rectum delineation, a seeded region-growing method was used. Manual contouring was considered the reference method. The following parameters were measured: volume ratio (Vr) (automatic/manual), volume overlap (Vo) (ratio of the volume of intersection to the volume of union; optimal value = 1), and correctly delineated volume (Vc) (percent ratio of the volume of intersection to the manually defined volume; optimal value = 100).

RESULTS

For the CTV, the Vr, Vo, and Vc were 1.13 (+/-0.1 SD), 0.78 (+/-0.05 SD), and 94.75 (+/-3.3 SD), respectively. For the rectum, the Vr, Vo, and Vc were 0.97 (+/-0.1 SD), 0.78 (+/-0.06 SD), and 86.52 (+/-5 SD), respectively. For the bladder, the Vr, Vo, and Vc were 0.95 (+/-0.03 SD), 0.88 (+/-0.03 SD), and 91.29 (+/-3.1 SD), respectively.

CONCLUSIONS

Our results show that the organ-model method is robust, and results in reproducible prostate segmentation with minor interactive corrections. For automatic bladder and rectum delineation, magnetic resonance imaging soft-tissue contrast enables the use of region-growing methods.

New developments in MRI for target volume delineation in radiotherapy

MRI is being increasingly used in oncology for staging, assessing tumour response and also for treatment planning in radiotherapy. Both conformal and intensity-modulated radiotherapy requires improved means of defining target volumes for treatment planning in order to achieve its intended benefits. MRI can add to the radiotherapy treatment planning (RTP) process by providing excellent and improved characterization of soft tissues compared with CT. Together with its multiplanar capability and increased imaging functionality, these advantages for target volume delineation outweigh its drawbacks of lacking electron density information and potential image distortion. Efficient MR distortion assessment and correction algorithms together with image co-registration and fusion programs can overcome these limitations and permit its use for RTP. MRI developments using new contrast media, such as ultrasmall superparamagnetic iron oxide particles for abnormal lymph node identification, techniques such as dynamic contrast enhanced MRI and diffusion MRI to better characterize tissue and tumour regions as well as ultrafast volumetric or cine MR sequences to define temporal patterns of target and organ at risk deformity and variations in spatial location have all increased the scope and utility of MRI for RTP. Information from these MR developments may permit treatment individualization, strategies of dose escalation and image-guided radiotherapy. These developments will be reviewed to assess their current and potential use for RTP and precision high dose radiotherapy.