Computer-assisted quantitative evaluation of therapeutic responses for lymphoma using serial PET/CT imaging

Development and evaluation of an articulated registration algorithm for human skeleton registration

Accurate registration over multiple scans is necessary to assess treatment response of bone diseases (e.g. metastatic bone lesions). This study aimed to develop and evaluate an articulated registration algorithm for the whole-body skeleton registration in human patients. In articulated registration, whole-body skeletons are registered by auto-segmenting into individual bones using atlas-based segmentation, and then rigidly aligning them. Sixteen patients (weight = 80-117 kg, height = 168-191 cm) with advanced prostate cancer underwent the pre- and mid-treatment PET/CT scans over a course of cancer therapy. Skeletons were extracted from the CT images by thresholding (HU>150). Skeletons were registered using the articulated, rigid, and deformable registration algorithms to account for position and postural variability between scans. The inter-observers agreement in the atlas creation, the agreement between the manually and atlas-based segmented bones, and the registration performances of all three registration algorithms were all assessed using the Dice similarity index-DSIobserved, DSIatlas, and DSIregister. Hausdorff distance (dHausdorff) of the registered skeletons was also used for registration evaluation. Nearly negligible inter-observers variability was found in the bone atlases creation as the DSIobserver was 96 ± 2%. Atlas-based and manual segmented bones were in excellent agreement with DSIatlas of 90 ± 3%. Articulated (DSIregsiter = 75 ± 2%, dHausdorff = 0.37 ± 0.08 cm) and deformable registration algorithms (DSIregister = 77 ± 3%, dHausdorff = 0.34 ± 0.08 cm) considerably outperformed the rigid registration algorithm (DSIregsiter = 59 ± 9%, dHausdorff = 0.69 ± 0.20 cm) in the skeleton registration as the rigid registration algorithm failed to capture the skeleton flexibility in the joints. Despite superior skeleton registration performance, deformable registration algorithm failed to preserve the local rigidity of bones as over 60% of the skeletons were deformed. Articulated registration is superior to rigid and deformable registrations by capturing global flexibility while preserving local rigidity inherent in skeleton registration. Therefore, articulated registration can be employed to accurately register the whole-body human skeletons, and it enables the treatment response assessment of various bone diseases.

Molecular MR imaging of neovascular progression in the Vx2 tumor with αvβ3-targeted paramagnetic nanoparticles

PURPOSE

To assess the dependence of neovascular molecular magnetic resonance (MR) imaging on relaxivity (r1) of αvβ3-targeted paramagnetic perfluorocarbon (PFC) nanoparticles and to delineate the temporal-spatial consistency of angiogenesis assessments for individual animals.

MATERIALS AND METHODS

Animal protocols were approved by the Washington University Animal Studies Committee. Proton longitudinal and transverse relaxation rates of αvβ3-targeted and nontargeted PFC nanoparticles incorporating gadolinium diethylenetrianime pentaacedic acid (Gd-DTPA) bisoleate (BOA) or gadolinium tetraazacyclododecane tetraacetic acid (Gd-DOTA) phosphatidylethanolamine (PE) into the surfactant were measured at 3.0 T. These paramagnetic nanoparticles were compared in 30 New Zealand White rabbits (four to six rabbits per group) 14 days after implantation of a Vx2 tumor. Subsequently, serial MR (3.0 T) neovascular maps were developed 8, 14, and 16 days after tumor implantation by using αvβ3-targeted Gd-DOTA-PE nanoparticles (n = 4) or nontargeted Gd-DOTA-PE nanoparticles (n = 4). Data were analyzed with analysis of variance and nonparametric statistics.

RESULTS

At 3.0 T, Gd-DTPA-BOA nanoparticles had an ionic r1 of 10.3 L · mmol(-1) · sec(-1) and a particulate r1 of 927000 L · mmol(-1) · sec(-1). Gd-DOTA-PE nanoparticles had an ionic r1 of 13.3 L · mmol(-1) · sec(-1) and a particulate r1 of 1 197000 L · mmol(-1) · sec(-1). Neovascular contrast enhancement in Vx2 tumors (at 14 days) was 5.4% ± 1.06 of the surface volume with αvβ3-targeted Gd-DOTA-PE nanoparticles and 3.0% ± 0.3 with αvβ3-targeted Gd-DTPA-BOA nanoparticles (P = .03). MR neovascular contrast maps of tumors 8, 14, and 16 days after implantation revealed temporally consistent and progressive surface enhancement (1.0% ± 0.3, 4.5% ± 0.9, and 9.3% ± 1.4, respectively; P = .0008), with similar time-dependent changes observed among individual animals.

CONCLUSION

Temporal-spatial patterns of angiogenesis for individual animals were followed to monitor longitudinal tumor progression. Neovasculature enhancement was dependent on the relaxivity of the targeted agent.

Semi-automatic analysis of standard uptake values in serial PET/CT studies in patients with lung cancer and lymphoma

BACKGROUND

Changes in maximum standardised uptake values (SUVmax) between serial PET/CT studies are used to determine disease progression or regression in oncologic patients. To measure these changes manually can be time consuming in a clinical routine. A semi-automatic method for calculation of SUVmax in serial PET/CT studies was developed and compared to a conventional manual method. The semi-automatic method first aligns the serial PET/CT studies based on the CT images. Thereafter, the reader selects an abnormal lesion in one of the PET studies. After this manual step, the program automatically detects the corresponding lesion in the other PET study, segments the two lesions and calculates the SUVmax in both studies as well as the difference between the SUVmax values. The results of the semi-automatic analysis were compared to that of a manual SUVmax analysis using a Philips PET/CT workstation. Three readers did the SUVmax readings in both methods. Sixteen patients with lung cancer or lymphoma who had undergone two PET/CT studies were included. There were a total of 26 lesions.

RESULTS

Linear regression analysis of changes in SUVmax show that intercepts and slopes are close to the line of identity for all readers (reader 1: intercept = 1.02, R2 = 0.96; reader 2: intercept = 0.97, R2 = 0.98; reader 3: intercept = 0.99, R2 = 0.98). Manual and semi-automatic method agreed in all cases whether SUVmax had increased or decreased between the serial studies. The average time to measure SUVmax changes in two serial PET/CT examinations was four to five times longer for the manual method compared to the semi-automatic method for all readers (reader 1: 53.7 vs. 10.5 s; reader 2: 27.3 vs. 6.9 s; reader 3: 47.5 vs. 9.5 s; p < 0.001 for all).

CONCLUSIONS

Good agreement was shown in assessment of SUVmax changes between manual and semi-automatic method. The semi-automatic analysis was four to five times faster to perform than the manual analysis. These findings show the feasibility of using semi-automatic methods for calculation of SUVmax in clinical routine and encourage further development of programs using this type of methods.

Automatic detection and classification of nasopharyngeal carcinoma on PET/CT with support vector machine

PURPOSE

Positron emission tomography/computed tomography (PET/CT) has established values for imaging of head and neck cancers, including the nasopharyngeal carcinoma (NPC), utilizing both morphologic and functional information. In this paper, we introduce a computerized system for automatic detection of NPC, targeting both the primary tumor and regional nodal metastasis, on PET/CT.

METHODS

Candidate lesions were extracted based on the features from both PET and CT images and a priori knowledge of anatomical features and subsequently classified by a support vector machine algorithm. The system was validated with 25 PET/CT examinations from 10 patients suffering from NPC. Lesions manually contoured by experienced radiologists were used as the gold standard.

RESULTS

Results showed that the system successfully identified all 53 hypermetabolic lesions larger than 1 cm in size and excluded normal physiological uptake in brown fat, muscles, bone marrow, brain, and salivary glands.

CONCLUSION

The system combined both imaging features and a priori clinical knowledge for classification between pathological and physiological uptake. Preliminary results showed that the system was highly accurate and promising for adoption in clinical use.