A novel supervised learning method to generate CT images for attenuation correction in delayed pet scans

Suggestions for topics suitable for these Point/Counterpoint debates should be addressed to Colin G. Orton, Professor Emeritus, Wayne State University, Detroit: ortonc@comcast.net. Persons participating in Point/Counterpoint discussions are selected for their knowledge and communicative skill. Their positions for or against a proposition may or may not reflect their personal opinions or the positions of their employers.

We evaluate the accuracy of scaling CT images for attenuation correction of PET data measured for bone. While the standard tri-linear approach has been well tested for soft tissues, the impact of CT-based attenuation correction on the accuracy of tracer uptake in bone has not been reported in detail. We measured the accuracy of attenuation coefficients of bovine femur segments and patient data using a tri-linear method applied to CT images obtained at different kVp settings. Attenuation values at 511 keV obtained with a 68Ga/68Ge transmission scan were used as a reference standard. The impact of inaccurate attenuation images on PET standardized uptake values (SUVs) was then evaluated using simulated emission images and emission images from five patients with elevated levels of FDG uptake in bone at disease sites. The CT-based linear attenuation images of the bovine femur segments underestimated the true values by 2.9 ± 0.3% for cancellous bone regardless of kVp. For compact bone the underestimation ranged from 1.3% at 140 kVp to 14.1% at 80 kVp. In the patient scans at 140 kVp the underestimation was approximately 2% averaged over all bony regions. The sensitivity analysis indicated that errors in PET SUVs in bone are approximately proportional to errors in the estimated attenuation coefficients for the same regions. The variability in SUV bias also increased approximately linearly with the error in linear attenuation coefficients. These results suggest that bias in bone uptake SUVs of PET tracers ranges from 2.4% to 5.9% when using CT scans at 140 and 120 kVp for attenuation correction. Lower kVp scans have the potential for considerably more error in dense bone. This bias is present in any PET tracer with bone uptake but may be clinically insignificant for many imaging tasks. However, errors from CT-based attenuation correction methods should be carefully evaluated if quantitation of tracer uptake in bone is important.

17113 Background: MPM is a rapidly progressive malignancy with a median survival of 6 to 9 months. Studies in non-small cell lung cancer, gastrointestinal stromal tumours, and breast cancer suggest that Positron Emission Tomography (PET) scanning may predict the clinical outcome of chemotherapy at an early stage of treatment. The role of PET in predicting and monitoring response to treatment in MPM is unclear. In this study we assessed if baseline and early interval PET scanning could be of benefit in MPM, correlating the results with standard Computerised Tomography (CT) assessment using the modified Response Evaluation Criteria In Solid Tumours (RECIST) of Byrne and Nowak. Methods: 10 patients with histologically proven MPM proceeded to baseline PET and CT scanning prior to chemotherapy. In 8 patients a second PET scan was performed within three weeks of commencing treatment. The following PET derived parameters were calculated: Standard Uptake Value (SUV)maximum, SUVaverage, PETvolume, and Total Lesion Glycolysis (TLG). Baseline SUVmaximum and early PET response was compared with CT response and survival. Results: Baseline SUVmaximum predicted subsequent CT response in the 10 patients studied. Four out of 5 patients with baseline SUVmaximum of <15 showed a subsequent partial response on CT, whilst the 5th patient had resolution of a pleural effusion. All 5 patients with baseline SUVmaximum of >15 had stable or progressive disease on CT. In contrast change in early SUVmaximum from baseline to 3 weeks did not predict subsequent CT response (Kappa value 0.07) in the 8 patients evaluated. Change in PET volume and TLG correlated with subsequent CT response (Kappa value= 0.80 for both). Only 1 patient with a partial response on % PETvolume and %TLG has died 14 months after commencing treatment. The other 3 patients are alive at 17+, 19+ and 20+ months. The 4 patients with stable or progressive disease using these parameters were deceased at 5, 6, 10, and 12 months respectively. Conclusion: PET may be useful in predicting and evaluating response to treatment in MPM. Furthermore, the results suggest that PET response may be a useful prognostic tool in this disease. It is important to use the appropriate PET derived parameters. Further study with a larger patient population is warranted. No significant financial relationships to disclose.

Small animal PET scanning with <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG is increasingly used in murine tumor model. Experimental condition of animal handling impacts on the biodistribution of <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG in mouse study. The aim of this study was to optimize experimental conditions of animal handling and registration method for imaging lung metastasis tumor by using <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG small animal PET and CT. The experimental conditions for <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG biodistribution and animal PET imaging were 20 plusmn 2 h fasting and warming kept at 30degC. Biodistribution, standard uptake value (SUV) image and glucose SUV (SUV <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">G</sub> ) of <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG were evaluated and blood glucose level was also measured. Female C57BL/6 mice were injected intravenously with 2 times 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sup> of B16-F10 melanoma cells which developed metastasis tumors in lung. In lung metastasis tumor model, <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG images was obtained in optimized anesthesia condition and registered with anatomical clinical CT image. The small animal PET and CT images were registered using the hardware fiducial markers and small animal contour point in both data sets to perform a point- based rigid registration. Isoflurane 0.5 condition was chosen for lung metastasis tumor imaging. Lung metastasis tumor was clearly visualized by <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG PET image with Iso 0.5 anesthesia. Tumor imaging in lung region with <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG small animal PET should be considered feeding, temperature, and especially choice and level of anesthetic for minimizing background radioactivity. Fused imaging with small animal PET and CT image could be useful for the detection of metastatic tumor. This method improves the quantitative accuracy and interpretation of the tracer.

The Case Against Heavy PETing

Attenuation correction of emission PET images with average CT: Interpolation from breath-hold CT

Recent advances in PET image reconstruction have focused on achieving high image quality and quantitative accuracy. Bayesian penalized likelihood (BPL) algorithms, such as Q.Clear and HYPER Iterative that have been integrated into commercial PET systems offer robust image noise suppression and edge preservation through regularization. In parallel, methods based on deep learning such as SubtlePET, AiCE, uAI® HYPER DLR, and Precision DL have emerged primarily as post-processing techniques. They use trained convolutional neural networks to reduce image noise while preserving lesion contrast. These methods have reduced image acquisition times or reduced radiotracer doses while maintaining diagnostic confidence. uAI® HYPER DPR represents a hybrid approach by embedding deep learning in iterative reconstruction. This review summarizes the technical principles and the clinical performance of BPL and deep learning-based PET reconstruction algorithms, and discusses key considerations such as image quality and quantitative accuracy of PET images. This review should deepen understanding of advanced PET image reconstruction techniques and accelerate their clinical implementation across diverse PET imaging applications.

Noninvasive Staging of Non-small Cell Lung Cancer: ACCP Evidenced-Based Clinical Practice Guidelines (2nd Edition)

The amount of radio-labeled (18)F-fluoro-2-deoxy-glucose (FDG) uptake, a measurement of the increased glucose metabolism of non-small-cell lung cancer (NSCLC) cells, has recently been correlated with proliferation capacity. The Standardized Uptake Value (SUV), a semi-quantitative measurement of FDG uptake on positron emission tomography (PET) scan, could thus be of prognostic significance. We analyzed the follow-up of 125 potentially operable NSCLC patients, previously included in three of our prospective PET protocols. Performance status, maximal tumor diameter, tumor-cell type, SUV, and final staging were analyzed for their possible association with survival. Sixty-five patients had stage I or II NSCLC, 37 had stage IIIA, and 23 had stage IIIB. Treatment was complete resection in 91 cases. In a univariate analysis, performance status (P =.002), stage (P =.001), tumor diameter (P =.06), tumor-cell type (P =.03), and SUV greater than 7 (P =.001) were correlated with survival. For SUV, group dichotomy with a cut-off SUV of 7 had the best discriminative value for prognosis, both in the total and surgical cohort. A multivariate Cox analysis identified performance status (P =.02), stage (P =.01), and SUV (P =.007) as important for the prognosis. In the surgical group, patients with a resected tumor less than 3 cm had an expected 2-year survival of 86%, if the SUV was below 7, and 60%, if above 7. Nearly all resected tumors larger than 3 cm had SUV's greater than 7 and an expected 2-year survival of 43%. We conclude that the FDG uptake in primary NSCLC on PET has an important prognostic value and could be complementary to other well-known factors in the decision on adjuvant treatment protocols.

Post-transplant lymphoproliferative disorder (PTLD) is a serious complication of transplantation. We examined the role of positron emission tomography (PET) scanning in PTLD. All patients treated for PTLD from 2001-2006 who also underwent PET scans were reviewed. Nineteen PTLD patients were included. Seventeen patients had PET scans for staging at diagnosis. Of these, two patients with primary central nervous system lymphoma and one patient with only bone marrow involvement after complete surgical resection of a bowel lesion had no abnormalities on CT or PET scan. The remaining patients had measurable, extracranial disease by CT scan and PET scan. The median maximum standard uptake value was 8.2 (range 3-30). Thirteen patients had a PET scan following treatment. Eleven of 13 patients had a complete response (CR). Two of 13 patients had persistent disease following therapy; in one of these patients, relapsed disease was documented by PET scan alone. Of the 11 patients with CR, three patients relapsed shortly thereafter. In each case, at the time of relapse, the PET scan confirmed recurrent disease regardless of histopathologic subtype. PET scans may have a role in the staging and follow-up of patients with PTLD. Additional prospective studies are warranted.

MP01-12 ACCURACY OF FDG-PET SCAN IN PRIMARY TESTICULAR SEMINOMA: ANALYSIS FROM SEMS TRIAL

A área da medicina nuclear possui uma crescente vertente em terapia de doenças, particularmente no tratamento de tumores radiosenssíveis. Devido à alta dose utilizada na terapia desses tumores é de extrema importância a quantificação da distribuição da dose, para assim evitar os efeitos deletérios nos tecidos sadios. No Brasil o sistema de dosimetria interna utilizado é o MIRD (Medical Internal Radiation Dose) baseado em um modelo de referência que não possui dados suficientes do paciente para obter uma avaliação detalhada da dose em terapia. Novos sistemas de dosimetria interna utilizam imagens de tomografia computadorizada para obter tanto informações do corpo do paciente, como informações da distribuição da atividade interna do paciente, para assim, com base nessas informações, realizar o cálculo de dose mediante um código de Monte Carlo. Esse tipo de dosimetria é denominado dosimetria personalisada do paciente. No Centro de Engenharia Nuclear do Ipen, está em desenvolvimento uma metodologia em que as informações das imagens tomográficas são inseridas no código de Monte Carlo MCNP5 mediante um software denominado SCMS (Sistema Construtor de Manequins Segmentadores). Assim, a dosimetria pode ser realizada de forma personalizada para cada paciente, obtendo-se a deposição de energia nos órgãos de interesse. O presente trabalho teve por objetivo contribuir para a área de medicina nuclear com o desenvolvimento de parte do sistema mencionado de dosimetria personalizada do paciente para terapia radionuclídica. Para isso foram propostos três objetivos específicos: (1)Desenvolver um software para converter imagens de tomografia computadorizada (CT) em parâmetros do tecido (&rho;,&omega;i); (2) Desenvolver um software para realizar a correção de atenuação em imagens tomográficas de medicina nuclear (SPECT ou PET) para fornecer o mapa de atividade relativa e (3) Preparar os dois softwares anteriores para fornecerem dados de entrada ao SCMS. O software desenvolvido para o primeiro objeto específico foi o Image Converter Computed Tomography (ICCT) que obteve uma boa precisão para determinar a densidade, e em relação à composição dos tecidos, os únicos elementos que obtiveram alta variação foram o carbono e oxigênio. Felizmente, essa variação para a faixa de energia utilizada em terapia radionuclídica não é prejudicial para a distribuição da dose. Uma das vantagens é a alta precisão em relação a cálcio e fósforo que possuem grande influência na distribuição da dose. Para o segundo objetivo foi desenvolvido o Attenuation Corretion PET SPECT (ACPS) que efetua a correção de atenuação em imagens de PET e SPECT mediante o método de Chang de 1a ordem e gera a distribuição da atividade relativa no interior do paciente. Por fim, os dados gerados pelos dois softwares foram formatados para o SCMS.

The Discovery ST PET-CT scanner corrects the attenuation of the 511 keV PET emission using an attenuation map derived from a CT scan which is acquired with a 50 cm detector field of view. The PET scanner uses a 70 cm reconstruction field of view, and to support attenuation correction over the 70 cm diameter, the CT data are extended using a previously published method. The restored CT data have some errors in the extrapolated outline and the density in the extrapolated region, but the integral attenuation along the PET lines of response are restored with an acceptable error for the purpose of emission attenuation correction. The effect of the truncation on the reconstructed PET activity and the reduction in these effects with the restoration algorithm was measured using an active PET phantom. The use of the truncated attenuation map resulted in a loss of activity outside the truncated field of view and also had an effect on the activity within the 50 cm field of view. The use of the extrapolated CT images for attenuation correction results in a considerable improvement in the accuracy of the reconstructed activity.

Noninvasive Staging of Non-small Cell Lung Cancer : A Review of the Current Evidence

The incidence of adenocarcinomas of the distal oesophagus (ADE) has dramatically increased in Western countries. The clinical importance of a FDG PET finding discordant with CT was determined in patients with locally advanced ADE. In addition, tumour standardized uptake values (SUV) were correlated with patient survival. 40 consecutive patients were analyzed retrospectively. All patients underwent an attenuation corrected FDG PET scan (neck, chest, abdomen) and contrast enhanced helical CT of the chest and abdomen. PET and CT scans were reviewed independently and concomitantly with respect to metastases in predefined lymph node sites and organs. Any discordance between PET and CT was assessed for clinical relevance. Clinical relevance was defined as a change in the overall therapeutic concept (curative vs. palliative). Follow-up imaging and histological evaluation served as the gold standard. Mean tumour SUVs were determined by 1.5 cm regions of interest placed over the tumour's maximum. When read independently from the CT scan FDG PET indicated a clinically relevant change in tumour stage in 9/40 patients (23%) and a non-relevant change in 11/40 patients (28%). PET was correct in 5/9 patients (56%) with clinically relevant discordances. In 4/9 patients PET was incorrect (3 false positive due to suspicion of M1-lymph nodes or lung metastases, 1 false negative in disseminated liver metastases). With concomitant reading, PET indicated a clinically relevant change in tumour stage in 6/40 patients (15%) and a non-relevant change in 5/40 patients (13%). PET was correct in 5/6 patients (83%) with clinically relevant discordances. The patient with disseminated liver disease remained the single false negative. Overall, the benefit from PET was based on its higher diagnostic accuracy at organ sites. Tumour SUV did not correlate with patient survival. About half of discordances between FDG PET and CT are clinically relevant. Concomitant reading of PET and CT is advisable as it reduces the overall rate of discordances and enhances the accuracy of PET in clinical relevant discordances (from 56% to 83%).