Prediction of Lymph Node Manifestations in Malignant Lymphoma: Significant Role of Volumetric Compared With Established Metric Lymph Node Analysis in Multislice Computed Tomography

Evaluation of mediastinal lymph node segmentation of heterogeneous CT data with full and weak supervision

Accurate lymph node size estimation is critical for staging cancer patients, initial therapeutic management, and assessing response to therapy. Current standard practice for quantifying lymph node size is based on a variety of criteria that use uni-directional or bi-directional measurements. Segmentation in 3D can provide more accurate evaluations of the lymph node size. Fully convolutional neural networks (FCNs) have achieved state-of-the-art results in segmentation for numerous medical imaging applications, including lymph node segmentation. Adoption of deep learning segmentation models in clinical trials often faces numerous challenges. These include lack of pixel-level ground truth annotations for training, generalizability of the models on unseen test domains due to the heterogeneity of test cases and variation of imaging parameters. In this paper, we studied and evaluated the performance of lymph node segmentation models on a dataset that was completely independent of the one used to create the models. We analyzed the generalizability of the models in the face of a heterogeneous dataset and assessed the potential effects of different disease conditions and imaging parameters. Furthermore, we systematically compared fully-supervised and weakly-supervised methods in this context. We evaluated the proposed methods using an independent dataset comprising 806 mediastinal lymph nodes from 540 unique patients. The results show that performance achieved on the independent test set is comparable to that on the training set. Furthermore, neither the underlying disease nor the heterogeneous imaging parameters impacted the performance of the models. Finally, the results indicate that our weakly-supervised method attains 90%- 91% of the performance achieved by the fully supervised training.

The Liver Tumor Segmentation Benchmark (LiTS)

In this work, we report the set-up and results of the Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 and the International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2017 and 2018. The image dataset is diverse and contains primary and secondary tumors with varied sizes and appearances with various lesion-to-background levels (hyper-/hypo-dense), created in collaboration with seven hospitals and research institutions. Seventy-five submitted liver and liver tumor segmentation algorithms were trained on a set of 131 computed tomography (CT) volumes and were tested on 70 unseen test images acquired from different patients. We found that not a single algorithm performed best for both liver and liver tumors in the three events. The best liver segmentation algorithm achieved a Dice score of 0.963, whereas, for tumor segmentation, the best algorithms achieved Dices scores of 0.674 (ISBI 2017), 0.702 (MICCAI 2017), and 0.739 (MICCAI 2018). Retrospectively, we performed additional analysis on liver tumor detection and revealed that not all top-performing segmentation algorithms worked well for tumor detection. The best liver tumor detection method achieved a lesion-wise recall of 0.458 (ISBI 2017), 0.515 (MICCAI 2017), and 0.554 (MICCAI 2018), indicating the need for further research. LiTS remains an active benchmark and resource for research, e.g., contributing the liver-related segmentation tasks in http://medicaldecathlon.com/. In addition, both data and online evaluation are accessible via https://competitions.codalab.org/competitions/17094.

Towards More Structure: Comparing TNM Staging Completeness and Processing Time of Text-Based Reports versus Fully Segmented and Annotated PET/CT Data of Non-Small-Cell Lung Cancer

Results of PET/CT examinations are communicated as text-based reports which are frequently not fully structured. Incomplete or missing staging information can be a significant source of staging and treatment errors. We compared standard text-based reports to a manual full 3D-segmentation-based approach with respect to TNM completeness and processing time. TNM information was extracted retrospectively from 395 reports. Moreover, the RIS time stamps of these reports were analyzed. 2995 lesions using a set of 41 classification labels (TNM features + location) were manually segmented on the corresponding image data. Information content and processing time of reports and segmentations were compared using descriptive statistics and modelling. The TNM/UICC stage was mentioned explicitly in only 6% (n=22) of the text-based reports. In 22% (n=86), information was incomplete, most frequently affecting T stage (19%, n=74), followed by N stage (6%, n=22) and M stage (2%, n=9). Full NSCLC-lesion segmentation required a median time of 13.3 min, while the median of the shortest estimator of the text-based reporting time (R1) was 18.1 min (p=0.01). Tumor stage (UICC I/II: 5.2 min, UICC III/IV: 20.3 min, p < 0.001), lesion size (p < 0.001), and lesion count (n=1: 4.4 min, n=12: 37.2 min, p < 0.001) correlated significantly with the segmentation time, but not with the estimators of text-based reporting time. Numerous text-based reports are lacking staging information. A segmentation-based reporting approach tailored to the staging task improves report quality with manageable processing time and helps to avoid erroneous therapy decisions based on incomplete reports. Furthermore, segmented data may be used for multimedia enhancement and automatization.

Role of CT Density in PET/CT-Based Assessment of Lymphoma

PURPOSE

In patients with Hodgkin (HL) and non-Hodgkin lymphoma (NHL), primary staging, as well as intermediate and late response assessment, is often performed by integrated 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG) positron emission tomography/X-ray computed tomography (PET/CT). The purpose of this analysis was to evaluate if findings in patients with histopathologically proven HL or NHL might correlate with semi-automated density measurements of target lesions (TLs) in the CT component of the integrated PET/CT examination.

PROCEDURES

After approval by the institutional review board, 176 lymph nodes (LN) in 90 PET/CT examinations of 90 patients were retrospectively analyzed (HL, 108 TLs out of 55 patients; NHL, 68 TLs out of 35 patients). PET/CT was performed for reasons of primary staging, response evaluation as interim PET, or as final examination after therapy, according to the clinical schedule. Analyses of TLs were performed on the basis of tracer uptake (SUV) 60 min after tracer injection and volumetric CT histogram analysis in non-contrast-enhanced CT.

RESULTS

All patients were diagnosed with HL or NHL in a pretreatment biopsy. Prior to therapy induction, staging of all patients was performed using contrast-enhanced CT of the neck to the pelvis, or by [¹⁸F]FDG PET/CT. Of the 176 TLs, 119 were classified as malignant, and 57 were benign. Malignant TLs had significantly higher CT density values compared to benign (p < 0.01).

CONCLUSION

Density measurements of TLs in patients with HL and NHL correlate with the dignity of TLs and might therefore serve as a complementary surrogate parameter for the differentiation between malignant and benign TLs. A possible density threshold in clinical routine might be a 20-Hounsfield units (HU) cutoff value to rule out benignancy in TLs that are above the 20-HU threshold.

Radiomic Analysis using Density Threshold for FDG-PET/CT-Based N-Staging in Lung Cancer Patients

PURPOSE

Mediastinal nodal (N)-staging done by integrated 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG) positron emission tomography/x-ray computed tomography (PET/CT) in lung cancer patients is not always accurate. In order to reduce the need for invasive staging procedures, additional surrogate parameters for the detection of malignant lymph node infiltration would be helpful. The purpose of this study was to evaluate if radiomic semi-automated density profiling in mediastinal lymph nodes can improve preclinical N-staging, irrespective of the specific lung cancer entity.

PROCEDURES

This retrospective study was approved by the institutional review board. Two hundred forty-eight histologically proven lymph nodes in 122 lung cancer patients were investigated. In malignantly infiltrated lymph nodes, the specific lung cancer entity was histologically classified; benign lymph nodes were histologically classified as benign. Non-contrast enhanced [¹⁸F]FDG-PET/CT was performed before surgery/biopsy. Lymph node analyses were performed on the basis of FDG uptake and volumetric CT histogram analysis for metric lymph node sampling.

RESULTS

Of the 248 lymph nodes, 118 were benign, 130 malignant. Malignant lymph nodes had a significantly higher median CT density (32.4 Hounsfield units (HU) (min 5.4/max 77.5 HU)) compared to benign lymph nodes (9.3 HU (min -49.5/max 60.4 HU, p < 0.05), irrespective of the histological subtype. The discrimination between different malignant tumour subtypes by means of volumetric density analysis failed. Irrespective of the malignant subtype, a possible cutoff value of 20 HU may help differentiate between benign and malignant lymph nodes.

CONCLUSION

Density measurements in unclear mediastinal and hilar lymph nodes with equivocal FDG uptake in PET might serve as a possible surrogate parameter for N-staging in lung cancer patients, irrespective of the specific lung cancer subtype. This could also help to find possible high yield targets in cases where invasive lymph node staging is necessary.

Relapsed or Refractory Lymphoblastic Lymphoma in Children: Results and Analysis of 23 Patients in the EORTC 58951 and the LMT96 Protocols

BACKGROUND

The treatment of children with T-cell lymphoblastic lymphoma (T-LBL) and precursor B-cell lymphoblastic lymphoma (pB-LBL) has improved during the last decades. However, patients with relapsed or refractory lymphomas still have a poor prognosis.

METHODS

We report the characteristics and evolution of T-LBL and pB-LBL relapses in two multicenter prospective studies (LMT 96, European Organization for Research and Treatment of Cancer 58951).

RESULTS

From 1997 to 2008, 194 patients were included in these studies (157 T-LBL; 37 pB-LBL); among them, 23 patients underwent relapse or progression (18 T-LBL and 5 pB-LBL). The median age was 7.7 years (range 1.4-16.3). The survival rate at 8 years was 8.7% (21 deaths). The median time from diagnosis to relapse was 9 months [1-69] and 11 months [1-45] for T-LBL and pB-LBL, respectively. Twenty-two patients received a second-line treatment but remission was achieved in only seven patients. In 10 patients, intensification with hematopoietic stem cell transplantation (HSCT) was performed and four of them had a second relapse. Two patients still alive had T-LBL, experienced relapses 15 and 69 months after diagnosis, and received HSCT. Relapse during the intensive phase and second-line treatment without HSCT were identified as risk factors for bad prognosis (P = 0.01).

CONCLUSIONS

The results of second-line treatment, including intensive chemotherapy and HSCT, show that salvage treatment is still disappointing in controlling refractory forms. Early identification of patients at high risk of relapse is mandatory, allowing earlier intensification. Valid prognostic parameters, such as biological markers, are needed. International cooperation is warranted to collect more data on these rare diagnoses.

Automatic liver tumor segmentation on computed tomography for patient treatment planning and monitoring

Segmentation of liver tumors from Computed Tomography (CT) and tumor burden analysis play an important role in the choice of therapeutic strategies for liver diseases and treatment monitoring. In this paper, a new segmentation method for liver tumors from contrast-enhanced CT imaging is proposed. As manual segmentation of tumors for liver treatment planning is both labor intensive and time-consuming, a highly accurate automatic tumor segmentation is desired. The proposed framework is fully automatic requiring no user interaction. The proposed segmentation evaluated on real-world clinical data from patients is based on a hybrid method integrating cuckoo optimization and fuzzy c-means algorithm with random walkers algorithm. The accuracy of the proposed method was validated using a clinical liver dataset containing one of the highest numbers of tumors utilized for liver tumor segmentation containing 127 tumors in total with further validation of the results by a consultant radiologist. The proposed method was able to achieve one of the highest accuracies reported in the literature for liver tumor segmentation compared to other segmentation methods with a mean overlap error of 22.78 % and dice similarity coefficient of 0.75 in 3Dircadb dataset and a mean overlap error of 15.61 % and dice similarity coefficient of 0.81 in MIDAS dataset. The proposed method was able to outperform most other tumor segmentation methods reported in the literature while representing an overlap error improvement of 6 % compared to one of the best performing automatic methods in the literature. The proposed framework was able to provide consistently accurate results considering the number of tumors and the variations in tumor contrast enhancements and tumor appearances while the tumor burden was estimated with a mean error of 0.84 % in 3Dircadb dataset.

Real-time lymphography by indocyanine green fluorescence: improved navigation for regional lymph node staging

OBJECTIVES

Lymphatic imaging is an important step for the identification of lymphonodal positive disease in solid malignancies. Various methods have been established to detect positive lymph nodes, but the available diagnostic tools leave some inherent drawbacks. The aim of this study was to validate the indocyanine green (ICG) guided approach for transcutaneous and transmesenterial navigation with accurate lymph vessel and node identification for regional lymph node staging in solid malignancies.

METHODS

After institutional review board approval, a planar fluorescence imaging system was applied for lymphography and lymph node detection using ICG. A total of 96 patients were recruited and subject to fluorescence navigation for axillary (n = 46), inguinal (n = 16), and mesenterial (n = 34) visualization to analyze technical and clinical feasibility of the method after regional lymph node dissection and the applicability.

RESULTS

Overall fluorescence imaging identified lymphatic vessels and the SLN in 92 out of 96 patients (detection rate: 96%) after a mean injection of 7 mg ICG. Sensitivity based on fluorescent emission of ICG navigation was 95.6% in 65 out of 68 patients with lymph node dissection. All solid tumors were feasible for fluorescence-guided navigation with a broad spectrum.

CONCLUSION

Fluorescence-guided real-time lymphography with navigation to regional lymph nodes enables accurate visualization for a broad spectrum of different solid tumors with potential lymphonodal spread. In addition, the technique can be applied for lymphography in non-malignant diseases. With reference to the broad application spectrum, institutional investment in camera equipment can be justified.

Volumetric assessment of lymph node metastases in patients with non-seminomatous germ cell tumours treated with chemotherapy

INTRODUCTION

We evaluate volumetry and RECIST (Response Evaluation Criteria In Solid Tumors) as methodologies for response after chemotherapy for non-seminomatous germ cell tumour with retroperitoneal lymph node metastases.

METHODS

We performed a retrospective analysis of non-seminomatous testicular tumours and concurrent retroperitoneal lymph node metastases, which received chemotherapy and had computed tomography scans before and after treatment. Volumetric analysis and RECIST criteria were used to calculate response rates. We included a new category (favourable response) for patients with response rates between <100% and >70%. We calculated the correlation between volumetric and RECIST criteria with histological and clinical variables.

RESULTS

In total, 18 patients met the inclusion criteria. Histopathologic analysis of orchiectomy showed teratoma in 55.5% of patients, and those without teratoma had predominantly embryonal carcinoma. The mean baseline volume of retroperitoneal metastases was 447 cc, the mean post-chemotherapy volume was 33.6 cc, and the response rate was 62.6%. According to RECIST criteria, the mean baseline diameter was 4.93 cm, the mean post-chemotherapy diameter was 2.39 cm, and the response rate was 42.4%. Large post-chemotherapy residual masses correlated in both classifications with teratoma. The response rate was associated with the need for surgical treatment and the volumetric classification correlated with the need for lymphadenectomy.

CONCLUSIONS

This study evaluated volumetry as a way to measure clinical response in lymph node metastases of non-seminomatous germ cell tumours. Volumetric analysis is the next step in the evaluation of response rate; its accuracy remains to be determined. Teratoma had greater residual masses and our classification correlated with the need for lymphadenectomy.

Comparison of manual and semi-automatic measuring techniques in MSCT scans of patients with lymphoma: a multicentre study

OBJECTIVES

Multicentre evaluation of the precision of semi-automatic 2D/3D measurements in comparison to manual, linear measurements of lymph nodes regarding their inter-observer variability in multi-slice CT (MSCT) of patients with lymphoma.

METHODS

MSCT data of 63 patients were interpreted before and after chemotherapy by one/two radiologists in five university hospitals. In 307 lymph nodes, short (SAD)/long (LAD) axis diameter and WHO area were determined manually and semi-automatically. Volume was solely calculated semi-automatically. To determine the precision of the individual parameters, a mean was calculated for every lymph node/parameter. Deviation of the measured parameters from this mean was evaluated separately. Statistical analysis entailed intraclass correlation coefficients (ICC) and Kruskal-Wallis tests.

RESULTS

Median relative deviations of semi-automatic parameters were smaller than deviations of manually assessed parameters, e.g. semi-automatic SAD 5.3 vs. manual 6.5 %. Median variations among different study sites were smaller if the measurement was conducted semi-automatically, e. g. manual LAD 5.7/4.2 % vs. semi-automatic 3.4/3.4 %. Semi-automatic volumetry was superior to the other parameters (2.8 %).

CONCLUSIONS

Semi-automatic determination of different lymph node parameters is (compared to manually assessed parameters) associated with a slightly greater precision and a marginally lower inter-observer variability. These results are with regard to the increasing mobility of patients among different medical centres and in relation to the quality management of multicentre trials of importance.

KEY POINTS

• In a multicentre setting, semi-automatic measurements are more accurate than manual assessments. • Lymph node volumetry outperforms all other semi-automatically and manually performed measurements. • Use of semi-automatic lymph node analyses can reduce the inter-observer variability.

Prognostic significance of novel ¹⁸F-FDG PET/CT defined tumour variables in patients with oesophageal cancer

PURPOSE

(18)F-fluorodeoxyglucose ((18)F-FDG) positron emission tomography (PET) combined with computed tomography (PET/CT) is now established as a routine staging investigation of oesophageal cancer (OC). The aim of the study was to determine the prognostic significance of PET/CT defined tumour variables including maximum standardised uptake value (SUVmax), tumour length (TL), metastatic length of disease (MLoD), metabolic tumour volume (MTV), total lesion glycolysis (TLG) and total local nodal metastasis count (PET/CT LNMC).

MATERIALS AND METHODS

103 pre-treatment OC patients (76 adenocarcinoma, 25 squamous cell carcinoma, 1 poorly differentiated and 1 neuroendocrine tumour) were staged using PET/CT. The prognostic value of the measured tumour variables were tested using log-rank analysis of the Kaplan-Meier method and Cox's proportional hazards method. Primary outcome measure was survival from diagnosis.

RESULTS

Univariate analysis showed all variables to have strong statistical significance in relation to survival. Multivariate analysis demonstrated three variables that were significantly and independently associated with survival; MLoD (HR 1.035, 95% CI 1.008-1.064, p=0.011), TLG (HR 1.002, 95% CI 1.000-1.003, p=0.018) and PET/CT LNMC (HR 0.048-0.633, 95% CI 0.005-2.725, p=0.015).

CONCLUSION

MLoD, TLG, and PET/CT LNMC are important prognostic indicators in OC. This is the first study to demonstrate an independent statistical association between TLG, MLoD and survival by multivariable analysis, and highlights the value of staging OC patients with PET/CT using functional tumour variables.

Segmentation-based partial volume correction for volume estimation of solid lesions in CT

In oncological chemotherapy monitoring, the change of a tumor's size is an important criterion for assessing cancer therapeutics. Measuring the volume of a tumor requires its delineation in 3-D. This is called segmentation, which is an intensively studied problem in medical image processing. However, simply counting the voxels within a binary segmentation result can lead to significant differences in the volume, if the lesion has been segmented slightly differently by various segmentation procedures or in different scans, for example due to the limited spatial resolution of computed tomography (CT) or partial volume effects. This variability limits the sensitivity of size measurements and thus of therapy response assessments and it can even lead to misclassifications. We present a fast, generic algorithm for measuring the volume of solid, compact tumors in CT that considers partial volume effects at the border of a given segmentation result. The algorithm is an extension of the segmentation-based partial volume analysis proposed by Kuhnigk for the volumetry of solid lung lesions , such that it can be applied to inhomogeneous lesions and lesions with inhomogeneous surroundings. Our generalized segmentation-based partial volume correction is based on a spatial subdivision of the segmentation result, from which the fraction of tumor for each voxel is computed. It has been evaluated on phantom data, 1516 lesion segmentation pairs (lung nodules, liver metastases and lymph nodes) as well as 1851 lung nodules from the LIDC-IDRI database. The evaluations of our algorithm show a more accurate estimation of the real volume and its ability to reduce inter- and intra-observer variability significantly for each entity. Overall, the variability (interquartile range) for phantom data is reduced by 49% ( p ≪ 0.001) and the variability between different readers is reduced by 28% ( p ≪ 0.001). The average computation time is 0.2 s.

Volumetric response classification in metastatic solid tumors on MSCT: initial results in a whole-body setting

PURPOSE

To examine technical parameters of measurement accuracy and differences in tumor response classification using RECIST 1.1 and volumetric assessment in three common metastasis types (lung nodules, liver lesions, lymph node metastasis) simultaneously.

MATERIALS AND METHODS

56 consecutive patients (32 female) aged 41-82 years with a wide range of metastatic solid tumors were examined with MSCT for baseline and follow up. Images were evaluated by three experienced radiologists using manual measurements and semi-automatic lesion segmentation. Institutional ethics review was obtained and all patients gave written informed consent. Data analysis comprised interobserver variability operationalized as coefficient of variation and categorical response classification according to RECIST 1.1 for both manual and volumetric measures. Continuous data were assessed for statistical significance with Wilcoxon signed-rank test and categorical data with Fleiss kappa.

RESULTS

Interobserver variability was 6.3% (IQR 4.6%) for manual and 4.1% (IQR 4.4%) for volumetrically obtained sum of relevant diameters (p<0.05, corrected). 4-8 patients' response to therapy was classified differently across observers by using volumetry compared to standard manual measurements. Fleiss kappa revealed no significant difference in categorical agreement of response classification between manual (0.7558) and volumetric (0.7623) measurements.

CONCLUSION

Under standard RECIST thresholds there was no advantage of volumetric compared to manual response evaluation. However volumetric assessment yielded significantly lower interobserver variability. This may allow narrower thresholds for volumetric response classification in the future.

Diagnosis of regional node metastases in lung cancer with computer-aided 3D measurement of the volume and CT-attenuation values of lymph nodes

RATIONALE AND OBJECTIVES

The aim of this study is to assess the usefulness of computer-aided three-dimensional (3D) measurement of volume and computed tomography (CT) attenuation values of nodes for diagnosing nodal metastases of lung cancer.

MATERIALS AND METHODS

We measured three diameters, their ratios, volume, and CT values in 3D images of 191 nodes (64 malignant; 162 of <1 cm in short diameter) in 26 consecutive patients who underwent contrast-enhanced, thin-section, multidetector row CT before surgery. We separately studied statistically significant factors in a group of all nodes and in another group of nodes of <1 cm in short diameter with logistic modeling and evaluated their diagnostic accuracy.

RESULTS

Significant factors were CT values (P < .001) and short diameter (P = .001) for the total node group, and CT values (P = .030) and 3D volume (P = .035) for the <1 cm node group. Optimal 83% accuracy was obtained with a criterion of short diameter of >7.4 mm and CT values of >103 Hounsfield unit (HU) for the total node group, whereas optimal 76% accuracy was obtained with a criterion of 3D volume of >1282 mm(3) or CT values of >103 HU for the <1 cm node group.

CONCLUSION

3D measurement may be useful for diagnosing nodal metastases.