Musib Siddique

Quantifying tumour heterogeneity in 18F-FDG PET/CT imaging by texture analysis

Abstract18F-Fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) is now routinely used in oncological imaging for diagnosis and staging and increasingly to determine early response to treatment, often employing semiquantitative measures of lesion activity such as the standardized uptake value (SUV). However, the ability to predict the behaviour of a tumour in terms of future therapy response or prognosis using SUVs from a baseline scan prior to treatment is limited. It is recognized that medical images contain more useful information than may be perceived with the naked eye, leading to the field of “radiomics” whereby additional features can be extracted by computational postprocessing techniques. In recent years, evidence has slowly accumulated showing that parameters obtained by texture analysis of radiological images, reflecting the underlying spatial variation and heterogeneity of voxel intensities within a tumour, may yield additional predictive and prognostic information. It is hoped that measurement of these textural features may allow better tissue characterization as well as better stratification of treatment in clinical trials, or individualization of future cancer treatment in the clinic, than is possible with current imaging biomarkers. In this review we focus on the literature describing the emerging methods of texture analysis in 18FDG PET/CT, as well as other imaging modalities, and how the measurement of spatial variation of voxel grey-scale intensity within an image may provide additional predictive and prognostic information, and postulate the underlying biological mechanisms.

Differential effects of teriparatide on regional bone formation using 18F-fluoride positron emission tomography

Teriparatide increases skeletal mass, bone turnover markers, and bone strength, but local effects on bone tissue may vary between skeletal sites. We used positron emission tomography (PET) to study 18F‐fluoride plasma clearance (Ki) at the spine and standardized uptake values (SUVs) at the spine, pelvis, total hip, and femoral shaft in 18 postmenopausal women with osteoporosis. Subjects underwent a 1‐hour dynamic scan of the lumbar spine and a 10‐minute static scan of the pelvis and femurs at baseline and after 6 months of treatment with 20 µg/day teriparatide. Blood samples were taken to derive the arterial input function and lumbar spine Ki values evaluated using a three‐compartment model. SUVs were calculated for the spine, pelvis, total hip, and femoral shaft. After 6 months treatment with teriparatide, spine Ki values increased by 24% (p = .0003), while other model parameters were unchanged except for the fraction of tracer going to bone mineral (k3/[k2 + k3]), which increased by 23% (p = .0006). In contrast to Ki, spine SUVs increased by only 3% (p = .84). The discrepancy between changes in Ki and SUVs was explained by a 20% decrease in 18F− plasma concentration. SUVs increased by 37% at the femoral shaft (p = .0019), 20% at the total hip (p = .032), and 11% at the pelvis (p = .070). Changes in bone turnover markers and BMD were consistent with previous trials. We conclude that the changes in bone formation rate during teriparatide treatment as measured by 18F− PET differ at different skeletal sites, with larger increases in cortical bone than at trabecular sites.

Radiomics in PET: principles and applications

Radiomics is an evolving field in which the extraction of large amounts of features from diagnostic medical images may be used to predict underlying molecular and genetic characteristics, thereby improving treatment response prediction and prognostication and potentially allowing personalisation of cancer treatment. There is increasing interest in extracting additional data from PET images, particularly novel features that describe the heterogeneity of voxel intensities, but a number of potential limitations need to be recognised and overcome. Nevertheless, some early data suggest that extraction of additional quantitative data may offer further predictive and prognostic information in individual patients.

Predicting Response to Neoadjuvant Chemotherapy with PET Imaging Using Convolutional Neural Networks.

Imaging of cancer with 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) has become a standard component of diagnosis and staging in oncology, and is becoming more important as a quantitative monitor of individual response to therapy. In this article we investigate the challenging problem of predicting a patient’s response to neoadjuvant chemotherapy from a single 18F-FDG PET scan taken prior to treatment. We take a “radiomics” approach whereby a large amount of quantitative features is automatically extracted from pretherapy PET images in order to build a comprehensive quantification of the tumor phenotype. While the dominant methodology relies on hand-crafted texture features, we explore the potential of automatically learning low- to high-level features directly from PET scans. We report on a study that compares the performance of two competing radiomics strategies: an approach based on state-of-the-art statistical classifiers using over 100 quantitative imaging descriptors, including texture features as well as standardized uptake values, and a convolutional neural network, 3S-CNN, trained directly from PET scans by taking sets of adjacent intra-tumor slices. Our experimental results, based on a sample of 107 patients with esophageal cancer, provide initial evidence that convolutional neural networks have the potential to extract PET imaging representations that are highly predictive of response to therapy. On this dataset, 3S-CNN achieves an average 80.7% sensitivity and 81.6% specificity in predicting non-responders, and outperforms other competing predictive models.

The Precision and Sensitivity of 18F-Fluoride PET for Measuring Regional Bone Metabolism: A Comparison of Quantification Methods

The planning of research studies requires an understanding of the minimum number of subjects required. The aim of this study was to evaluate different methods of analyzing 18F-fluoride PET (18F− PET) dynamic spine scans to find the approach that requires the smallest sample size to detect a statistically significant response to treatment. Methods: Eight different approaches to 18F− PET analysis (3 variants of the Hawkins 3-tissue compartmental model, 3 variants of spectral analysis, deconvolution, and Patlak analysis) were used to evaluate the fluoride plasma clearance to bone mineral (Ki). Standardized uptake values (SUVs) were also studied. Data for 20 women who had 18F− PET spine scans at 0, 6, and 12 mo after stopping long-term bisphosphonate treatment were used to compare precision errors. Data for 18 women who had scans at baseline and 6 mo after starting teriparatide treatment were used to compare response to treatment. Results: The 4 approaches that fitted the rate constant k4 describing the reverse flow of 18F from bone as a free variable showed close agreement in Ki values, with correlation coefficients greater than 0.97. Their %CVs were 14.4%–14.8%, and treatment response to teriparatide was 23.2%–23.8%. The 3 methods that assumed k4 = 0 gave Ki values 20%–25% lower than the other methods, with correlation coefficients of 0.83–0.94, percentage coefficients of variation (%CVs) of 12.9%–13.3%, and treatment response of 25.2%–28.3%. A Hawkins model with k4 = 0.01 min−1 did not perform any better (%CV, 14.2%; treatment response, 26.1%). Correlation coefficients between SUV and the different Ki methods varied between 0.60 and 0.65. Although SUV gave the best precision (%CV, 10.1%), the treatment response (3.1%) was not statistically significant. Conclusion: Methods that calculated Ki assuming k4 = 0 required fewer subjects to demonstrate a statistically significant response to treatment than methods that fitted k4 as a free variable. Although SUV gave the smallest precision error, the absence of any significant changes make it unsuitable for examining response to treatment in this study.

Radionuclide studies of bone metabolism: do bone uptake and bone plasma clearance provide equivalent measurements of bone turnover?

Quantitative radionuclide imaging using (18)F-fluoride positron emission tomography (18F-PET) or (99m)Tc-methylene diphosphonate ((99m)Tc-MDP) bone scans provides a novel tool for studying regional and whole skeleton bone turnover that complements the information provided by biochemical markers. Radionuclide bone scans can be quantified by measuring either tracer uptake or, if blood sampling is performed, bone plasma clearance. This study examines whether these two methods provide equivalent information about bone turnover. We examined data from two clinical trials of the bone anabolic agent teriparatide. In Study 1 twenty osteoporotic women had 18F-PET scans of the lumbar spine at baseline and after 6 months treatment with teriparatide. Bone uptake in the lumbar spine was expressed as standardised uptake values (SUV) and blood samples taken to evaluate plasma clearance. In Study 2 ten women had (99m)Tc-MDP scans at baseline, 3 and 18 months after starting teriparatide. Blood samples were taken and whole skeleton plasma clearance and bone uptake calculated. In Study 1 spine plasma clearance increased by 23.8% after 6-months treatment (P=0.0003), whilst SUV increased by only 3.0% (P=0.84). In Study 2 whole skeleton plasma clearance increased by 37.1% after 18-months treatment (P=0.0002), whilst the 4-hour whole skeleton uptake increased by only 25.5% (P=0.0001). During treatment the 18F- plasma concentration decrease by 20% and (99m)Tc-MDP concentration by 13%, and these latter changes were sufficient to explain the differences between the uptake and plasma clearance results. Measurements of response to treatment using bone uptake and plasma clearance gave different results because the effects of teriparatide on bone resulted in a sufficiently increased demand for radionuclide tracer from the skeleton that the concentration in the circulation decreased. Similar effects may occur with other therapies that have a large enough effect on bone metabolism. In these circumstances changes in bone plasma clearance give a truer impression of response to treatment than those in SUV or uptake.

18F‐fluoride PET as a noninvasive imaging biomarker for determining treatment efficacy of bone active agents at the hip: A prospective, randomized, controlled clinical study

The functional imaging technique of 18F‐fluoride positron emission tomography (18F‐PET) allows the noninvasive quantitative assessment of regional bone formation at any skeletal site, including the spine and hip. The aim of this study was to determine if 18F‐PET can be used as an early biomarker of treatment efficacy at the hip. Twenty‐seven treatment‐naive postmenopausal women with osteopenia were randomized to receive teriparatide and calcium and vitamin D (TPT group, n = 13) or calcium and vitamin D only (control group, n = 14). Subjects in the TPT group were treated with 20 µg/day teriparatide for 12 weeks. 18F‐PET scans of the proximal femur, pelvis, and lumbar spine were performed at baseline and 12 weeks. The plasma clearance of 18F‐fluoride to bone, Ki, a validated measurement of bone formation, was measured at four regions of the hip, lumbar spine, and pelvis. A significant increase in Ki was observed at all regions of interest (ROIs), including the total hip (+27%, p = 0.002), femoral neck (+25%, p = 0.040), hip trabecular ROI (+21%, p = 0.017), and hip cortical ROI (+51%, p = 0.001) in the TPT group. Significant increases in Ki in response to TPT were also observed at the lumbar spine (+18%, p = 0.001) and pelvis (+42%, p = 0.001). No significant changes in Ki were observed for the control group. Changes in BMD and bone turnover markers were consistent with previous trials of teriparatide. In conclusion, this is the first study to our knowledge to demonstrate that 18F‐PET can be used as an imaging biomarker for determining treatment efficacy at the hip as early as 12 weeks after initiation of therapy.

¹⁸F-fluoride PET as a non-invasive imaging biomarker for determining treatment efficacy of bone active agents at the hip

The functional imaging technique of 18F‐fluoride positron emission tomography (18F‐PET) allows the noninvasive quantitative assessment of regional bone formation at any skeletal site, including the spine and hip. The aim of this study was to determine if 18F‐PET can be used as an early biomarker of treatment efficacy at the hip. Twenty‐seven treatment‐naive postmenopausal women with osteopenia were randomized to receive teriparatide and calcium and vitamin D (TPT group, n = 13) or calcium and vitamin D only (control group, n = 14). Subjects in the TPT group were treated with 20 µg/day teriparatide for 12 weeks. 18F‐PET scans of the proximal femur, pelvis, and lumbar spine were performed at baseline and 12 weeks. The plasma clearance of 18F‐fluoride to bone, Ki, a validated measurement of bone formation, was measured at four regions of the hip, lumbar spine, and pelvis. A significant increase in Ki was observed at all regions of interest (ROIs), including the total hip (+27%, p = 0.002), femoral neck (+25%, p = 0.040), hip trabecular ROI (+21%, p = 0.017), and hip cortical ROI (+51%, p = 0.001) in the TPT group. Significant increases in Ki in response to TPT were also observed at the lumbar spine (+18%, p = 0.001) and pelvis (+42%, p = 0.001). No significant changes in Ki were observed for the control group. Changes in BMD and bone turnover markers were consistent with previous trials of teriparatide. In conclusion, this is the first study to our knowledge to demonstrate that 18F‐PET can be used as an imaging biomarker for determining treatment efficacy at the hip as early as 12 weeks after initiation of therapy.

Radionuclide and Hybrid Bone Imaging

Quantitative bone scan imaging has a useful role in studies of the pathophysiology of metabolic bone disease and the response of patients to treatment. The advantage of nuclear medicine imaging as a way of studying bone remodelling is that it offers a unique way of measuring bone turnover both for the whole skeleton and in selected localised regions of interest (ROI). This chapter reviews methods of quantifying 99mTc-MDP and 18F-fluoride skeletal tracer kinetics by combining imaging data with blood sampling to measure bone plasma clearance. For studies using 99mTc-MDP, we describe three methods of measuring whole-skeleton plasma clearance (K bone): (1) The area-under-the-curve (AUC) method based on taking six blood samples between 5 min and 4 h and measuring the plasma concentration of free 99mTc-MDP by ultrafiltration. The AUC method requires a simultaneous measurement of glomerular filtration rate (GFR) using 51Cr-EDTA as a co-tracer. (2) The modified Brenner method, which measures K bone by drawing a soft tissue ROI over the adductor muscles and plotting the soft tissue counts at 1, 2, 3 and 4 h against the AUC values at the corresponding time points. (3) The Patlak method based on combining gamma camera measurements of whole-body retention with plasma data and finding K bone from the slope of the Patlak plot fitted to the 2, 3 and 4 h data points. Unlike the first two methods, the Patlak plot can also be used to measure regional values of K bone for any chosen ROI. Studies of 18F-fluoride skeletal tracer kinetics are performed using a 60-min dynamic positron emission tomography (PET) scan with measurement of the input function by either direct arterial sampling or using an image-derived input function from the heart, aorta or femoral artery and can be used to measure the net plasma clearance to the bone mineral compartment (K i ) at either the spine, hip or humerus. Further studies are required comparing radionuclide measurements with the gold standard of bone biopsy. Nuclear medicine measurements of bone turnover have an established role as a research technique, and there is a need for further studies to examine their role in assessing the pathophysiology of metabolic bone diseases, such as osteoporosis, Paget’s disease and renal osteodystrophy, and better understanding the effects of new pharmaceutical treatments at various sites throughout the skeleton.

Validation of image-derived arterial input functions at the femoral artery using 18F-fluoride positron emission tomography

IntroductionThe use of image-derived arterial input functions (IDAIF) for the dynamic quantification of bone metabolism using 18F-fluoride positron emission tomography 18F-PET is an attractive alternative to direct arterial blood sampling. Purposes(a) To validate a method for obtaining the IDAIF by imaging the femoral artery against a method for deriving the IDAIF at the aorta that was previously validated against direct arterial sampling. (b) To compare the accuracy of bone plasma clearance measurements (Ki) at the total hip site obtained using the femoral artery IDAIF against Ki values at the same site obtained using the aorta IDAIF. MethodsTwelve healthy postmenopausal women with a mean age of 62.6 years (range, 52.3–70.6 years) had 60-min dynamic 18F-PET scans of the lumbar spine and proximal femur 2 weeks apart. The femoral artery IDAIF was obtained from the proximal femur scan using four different algorithms: (a) fixed partial volume correction (PVC) method; (b) variable PVC method; (c) Chen method; and (d) Cook–Lodge method. The aorta IDAIF was obtained from the lumbar spine scan using a previously validated method and the respective Ki values in the hip were used to assess the performance of each of the femoral artery algorithms. ResultsWhen the femoral artery IDAIF methods were compared with the aorta IDAIF in terms of the area under the curve AUC values calculated in 4-min time intervals over 0–60 min, the absolute root mean square errors were: (a) fixed PVC, 0.52; (b) variable PVC, 0.54; (c) Chen, 0.72; and (d) Cook–Lodge, 0.49 in MBq s/ml. There were small, but statistically significant differences, in the Ki values found by all four femoral artery IDAIF methods when compared with the figures obtained using the aorta IDAIF. Bland–Altman plots of Ki values showed the best agreement for the fixed PVC method with a standard deviation of 0.0020 ml/min/ml, followed by variable PVC, Cook–Lodge and Chen method with standard deviations of 0.0022, 0.0024 and 0.0042 ml/min/ml, respectively. ConclusionWe have demonstrated that it is possible to measure regional bone turnover at the hip without the need to perform direct arterial sampling to acquire the arterial input function (AIF). The differences in the Ki values obtained at the hip by using aorta IDAIF and any of the four image-based AIF methods at the femoral artery were small and clinically insignificant. The performance of fixed PVC, variable PVC and Cook–Lodge method was similar although the latter was less robust than the other two methods.