An iterative technique to segment PET lesions using a Monte Carlo based mathematical model.

PURPOSE: The need for an accurate lesion segmentation tool in 18FDG PET is a prerequisite for the estimation of lesion response to therapy, for radionuclide dosimetry, and for the application of 18FDG PET to radiotherapy planning. In this work, the authors have developed an iterative method based on a mathematical fit deduced from Monte Carlo simulations to estimate tumor segmentation thresholds. METHODS: The GATE software, a GEANT4 based Monte Carlo tool, was used to model the GE Advance PET scanner geometry. Spheres ranging between 1 and 6 cm in diameters were simulated in a 10 cm high and 11 cm in diameter cylinder. The spheres were filled with water-equivalent density and simulated in both water and lung equivalent background. The simulations were performed with an infinite, 8/1, and 4/1 target-to-background ratio (T/B). A mathematical fit describing the correlation between the lesion volume and the corresponding optimum threshold value was then deduced through analysis of the reconstructed images. An iterative method, based on this mathematical fit, was developed to determine the optimum threshold value. The effects of the lesion volume and T/B on the threshold value were investigated. This method was evaluated experimentally using the NEMA NU2-2001 IEC phantom, the ACNP cardiac phantom, a randomly deformed aluminum can, and a spheroidal shape phantom implemented artificially in the lung, liver, and brain of patient PET images. Clinically, the algorithm was evaluated in six lesions from five patients. Clinical results were compared to CT volumes. RESULTS: This mathematical fit predicts an existing relationship between the PET lesion size and the percent of maximum activity concentration within the target volume (or threshold). It also showed a dependence of the threshold value on the T/B, which could be eliminated by background subtraction. In the phantom studies, the volumes of the segmented PET targets in the PET images were within 10% of the nominal ones. Clinically, the PET target volumes were also within 10% of those measured from CT images. CONCLUSIONS: This iterative algorithm enabled accurately segment PET lesions, independently of their contrast value.

Phase and amplitude binning for 4D-CT imaging.

We compare the consistency and accuracy of two image binning approaches used in 4D-CT imaging. One approach, phase binning (PB), assigns each breathing cycle 2pi rad, within which the images are grouped. In amplitude binning (AB), the images are assigned bins according to the breathing signal's full amplitude. To quantitate both approaches we used a NEMA NU2-2001 IEC phantom oscillating in the axial direction and at random frequencies and amplitudes, approximately simulating a patient's breathing. 4D-CT images were obtained using a four-slice GE Lightspeed CT scanner operating in cine mode. We define consistency error as a measure of ability to correctly bin over repeated cycles in the same field of view. Average consistency error mue+/-sigmae in PB ranged from 18%+/-20% to 30%+/-35%, while in AB the error ranged from 11%+/-14% to 20%+/-24%. In PB nearly all bins contained sphere slices. AB was more accurate, revealing empty bins where no sphere slices existed. As a proof of principle, we present examples of two non-small cell lung carcinoma patients' 4D-CT lung images binned by both approaches. While AB can lead to gaps in the coronal images, depending on the patient's breathing pattern, PB exhibits no gaps but suffers visible artifacts due to misbinning, yielding images that cover a relatively large amplitude range. AB was more consistent, though often resulted in gaps when no data existed due to patients' breathing pattern. We conclude AB is more accurate than PB. This has important consequences to treatment planning and diagnosis.

Evaluation of an automated deformable image matching method for quantifying lung motion in respiration-correlated CT images.

We have evaluated an automated registration procedure for predicting tumor and lung deformation based on CT images of the thorax obtained at different respiration phases. The method uses a viscous fluid model of tissue deformation to map voxels from one CT dataset to another. To validate the deformable matching algorithm we used a respiration-correlated CT protocol to acquire images at different phases of the respiratory cycle for six patients with nonsmall cell lung carcinoma. The position and shape of the deformable gross tumor volumes (GTV) at the end-inhale (EI) phase predicted by the algorithm was compared to those drawn by four observers. To minimize interobserver differences, all observers used the contours drawn by a single observer at end-exhale (EE) phase as a guideline to outline GTV contours at EI. The differences between model-predicted and observer-drawn GTV surfaces at EI, as well as differences between structures delineated by observers at EI (interobserver variations) were evaluated using a contour comparison algorithm written for this purpose, which determined the distance between the two surfaces along different directions. The mean and 90% confidence interval for model-predicted versus observer-drawn GTV surface differences over all patients and all directions were 2.6 and 5.1 mm, respectively, whereas the mean and 90% confidence interval for interobserver differences were 2.1 and 3.7 mm. We have also evaluated the algorithm's ability to predict normal tissue deformations by examining the three-dimensional (3-D) vector displacement of 41 landmarks placed by each observer at bronchial and vascular branch points in the lung between the EE and EI image sets (mean and 90% confidence interval displacements of 11.7 and 25.1 mm, respectively). The mean and 90% confidence interval discrepancy between model-predicted and observer-determined landmark displacements over all patients were 2.9 and 7.3 mm, whereas interobserver discrepancies were 2.8 and 6.0 mm. Paired t tests indicate no significant statistical differences between model predicted and observer drawn structures. We conclude that the accuracy of the algorithm to map lung anatomy in CT images at different respiratory phases is comparable to the variability in manual delineation. This method has therefore the potential for predicting and quantifying respiration-induced tumor motion in the lung.

Quantitation of respiratory motion during 4D-PET/CT acquisition.

We report on the variability of the respiratory motion during 4D-PET/CT acquisition. The respiratory motion for five lung cancer patients was monitored by tracking external markers placed on the abdomen. CT data were acquired over an entire respiratory cycle at each couch position. The x-ray tube status was recorded by the tracking system, for retrospective sorting of the CT data as a function of respiration phase. Each respiratory cycle was sampled in ten equal bins. 4D-PET data were acquired in gated mode, where each breathing cycle was divided into ten 500 ms bins. For both CT and PET acquisition, patients received audio prompting to regularize breathing. The 4D-CT and 4D-PET data were then correlated according to their respiratory phases. The respiratory periods, and average amplitude within each phase bin, acquired in both modality sessions were then analyzed. The average respiratory motion period during 4D-CT was within 18% from that in the 4D-PET sessions. This would reflect up to 1.8% fluctuation in the duration of each 4D-CT bin. This small uncertainty enabled good correlation between CT and PET data, on a phase-to-phase basis. Comparison of the average-amplitude within the respiration trace, between 4D-CT and 4D- PET, on a bin-by-bin basis show a maximum deviation of approximately 15%. This study has proved the feasibility of performing 4D-PET/CT acquisition. Respiratory motion was in most cases consistent between PET and CT sessions, thereby improving both the attenuation correction of PET images, and co-registration of PET and CT images. On the other hand, in two patients, there was an increased partial irregularity in their breathing motion, which would prevent accurately correlating the corresponding PET and CT images.

Four-dimensional (4D) PET/CT imaging of the thorax.

We have reported in our previous studies on the methodology, and feasibility of 4D-PET (Gated PET) acquisition, to reduce respiratory motion artifact in PET imaging of the thorax. In this study, we expand our investigation to address the problem of respiration motion in PET/CT imaging. The respiratory motion of four lung cancer patients were monitored by tracking external markers placed on the thorax. A 4D-CT acquisition was performed using a "step-and-shoot" technique, in which computed tomography (CT) projection data were acquired over a complete respiratory cycle at each couch position. The period of each CT acquisition segment was time stamped with an "x-ray ON" signal, which was recorded by the tracking system. 4D-CT data were then sorted into 10 groups, according to their corresponding phase of the breathing cycle. 4D-PET data were acquired in the gated mode, where each breathing cycle was divided into ten 0.5 s bins. For both CT and PET acquisitions, patients received audio prompting to regularize breathing. The 4D-CT and 4D-PET data were then correlated according to respiratory phase. The effect of 4D acquisition on improving the co-registration of PET and CT images, reducing motion smearing, and consequently increase the quantitation of the SUV, were investigated. Also, quantitation of the tumor motions in PET, and CT, were studied and compared. 4D-PET with matching phase 4D-CTAC showed an improved accuracy in PET-CT image co-registration of up to 41%, compared to measurements from 4D-PET with clinical-CTAC. Gating PET data in correlation with respiratory motion reduced motion-induced smearing, thereby decreasing the observed tumor volume, by as much as 43%. 4D-PET lesions volumes showed a maximum deviation of 19% between clinical CT and phase- matched 4D-CT attenuation corrected PET images. In CT, 4D acquisition resulted in increasing the tumor volume in two patients by up to 79%, and decreasing it in the other two by up to 35%. Consequently, these corrections have yielded an increase in the measured SUV by up to 16% over the clinical measured SUV, and 36% over SUV's measured in 4D-PET with clinical-CT Attenuation Correction (CTAC) SUV's. Quantitation of the maximum tumor motion amplitude, using 4D-PET and 4D-CT, showed up to 30% discrepancy between the two modalities. We have shown that 4D PET/CT is clinically a feasible method, to correct for respiratory motion artifacts in PET/CT imaging of the thorax. 4D PET/CT acquisition can reduce smearing, improve the accuracy in PET-CT co-registration, and increase the measured SUV. This should result in an improved tumor assessment for patients with lung malignancies.

Effect of respiratory gating on reducing lung motion artifacts in PET imaging of lung cancer.

Positron emission tomography (PET) has shown an increase in both sensitivity and specificity over computed tomography (CT) in lung cancer. However, motion artifacts in the 18F fluorodioxydoglucose (FDG) PET images caused by respiration persists to be an important factor in degrading PET image quality and quantification. Motion artifacts lead to two major effects: First, it affects the accuracy of quantitation, producing a reduction of the measured standard uptake value (SUV). Second, the apparent lesion volume is overestimated. Both impact upon the usage of PET images for radiation treatment planning. The first affects the visibility, or contrast, of the lesion. The second results in an increase in the planning target volume, and consequently a greater radiation dose to the normal tissues. One way to compensate for this effect is by applying a multiple-frame capture technique. The PET data are then acquired in synchronization with the respiratory motion. Reduction in smearing due to gating was investigated in both phantoms and patient studies. Phantom studies showed a dependence of the reduction in smearing on the lesion size, the motion amplitude, and the number of bins used for data acquisition. These studies also showed an improvement in the target-to-background ratio, and a more accurate measurement of the SUV. When applied to one patient, respiratory gating showed a 28% reduction in the total lesion volume, and a 56.5% increase in the SUV. This study was conducted as a proof of principle that a gating technique can effectively reduce motion artifacts in PET image acquisition.

Standardized uptake value in pediatric patients: an investigation to determine the optimum measurement parameter.

Although the standardized uptake value (SUV) is currently used in fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) imaging, concerns have been raised over its accuracy and clinical relevance. Dependence of the SUV on body weight has been observed in adults and this should be of concern in the pediatric population, since there are significant body changes during childhood. The aim of the present study was to compare SUV measurements based on body weight, body surface area and lean body mass in the pediatric population and to determine a more reliable parameter across all ages. Sixty-eight pediatric FDG-PET studies were evaluated. Age ranged from 2 to 17 years and weight from 11 to 77 kg. Regions of interest were drawn at the liver for physiologic comparison and at FDG-avid malignant lesions. SUV based on body weight (SUV(bw)) varied across different weights, a phenomenon less evident when body surface area (SUV(bsa)) normalization is applied. Lean body mass-based SUV (SUV(lbm)) also showed a positive correlation with weight, which again was less evident when normalized to bsa (SUV(bsa-lbm)). The measured liver SUV(bw) was 1.1+/-0.3, a much lower value than in our adult population (1.9+/-0.3). The liver SUV(bsa) was 7.3+/-1.3. The tumor sites had an SUV(bw) of 4.0+/-2.7 and an SUV(bsa) of 25.9+/-15.4 (65% of the patients had neuroblastoma). The bsa-based SUVs were more constant across the pediatric ages and were less dependent on body weight than the SUV(bw). These results indicate that SUV calculated on the basis of body surface area is a more uniform parameter than SUV based on body weight in pediatric patients and is probably the most appropriate approach for the follow-up of these patients.

FDG-PET standardized uptake values in normal anatomical structures using iterative reconstruction segmented attenuation correction and filtered back-projection.

Filtered back-projection (FBP) is the most commonly used reconstruction method for PET images, which are usually noisy. The iterative reconstruction segmented attenuation correction (IRSAC) algorithm improves image quality without reducing image resolution. The standardized uptake value (SUV) is the most clinically utilized quantitative parameter of [fluorine-18]fluoro-2-deoxy-D-glucose (FDG) accumulation. The objective of this study was to obtain a table of SUVs for several normal anatomical structures from both routinely used FBP and IRSAC reconstructed images and to compare the data obtained with both methods. Twenty whole-body PET scans performed in consecutive patients with proven or suspected non-small cell lung cancer were retrospectively analyzed. Images were processed using both IRSAC and FBP algorithms. Nonquantitative or gaussian filters were used to smooth the transmission scan when using FBP or IRSAC algorithms, respectively. A phantom study was performed to evaluate the effect of different filters on SUV. Maximum and average SUVs (SUVmax and SUVavg) were calculated in 28 normal anatomical structures and in one pathological site. The phantom study showed that the use of a nonquantitative smoothing filter in the transmission scan results in a less accurate quantification and in a 20% underestimation of the actual measurement. Most anatomical structures were identified in all patients using the IRSAC images. On average, SUVavg and SUVmax measured on IRSAC images using a gaussian filter in the transmission scan were respectively 20% and 8% higher than the SUVs calculated from conventional FBP images. Scatterplots of the data values showed an overall strong relationship between IRSAC and FBP SUVs. Individual scatterplots of each site demonstrated a weaker relationship for lower SUVs and for SUVmax than for higher SUVs and SUVavg. A set of reference values was obtained for SUVmax and SUVavg of normal anatomical structures, calculated with both IRSAC and FBP image reconstruction algorithms. The use of IRSAC and a gaussian filter for the transmission scan seems to give more accurate SUVs than are obtained from conventional FBP images using a nonquantitative filter for the transmission scan.

Whole-Body FDG-PET in Patients with Recurrent Colorectal Carcinoma. A Comparative Study with CT.

Purpose: To assess the clinical accuracy of whole-body 2-[F-18]-fluoro-2-deoxy-D-glucose-positron emission tomography (FDG-PET) in the diagnosis of recurrent colorectal carcinoma in comparison to conventional computed tomography (CT).Materials and methods: Forty patients with suspected recurrent colorectal carcinoma based on either progressive serial carcinoemrbyonic antigen (CEA) serum elevation or positive/equivocal CT findings underwent whole-body FDG-PET. PET results were compared with those of CT and correlated to the final histopathological and clinical findings.Results: A final diagnosis was obtained at 93 sites in 35 patients by histology and in 5 patients by clinical follow up of at least 6 months. Of the 93 sites, 53 were determined to be malignant and 40 benign. FDG-PET evaluated on a 5-point scale (0-4) showed a positive and negative predictive value in the range of 96-98% and 83-93% respectively as the threshold for positivity was moved from 0 through 3. By comparison, CT, also evaluated on a 5-point scale showed a positive and negative predictive value in the range of 75-88% and 67-71% respectively. The area under the fitted receiver operating characteristic curve for PET: A(PET) = 0.96 +/- 0.02 was significantly greater (P < 0.001) than that observed for CT: A(CT) = 0.77 +/- 0.06. The distribution of maximum standardized uptake value (SUVmax) showed that all negative lesions have SUVmax below 5.0 whereas 75% of positive lesions were above 5.0 pointing to the fact that disease positivity is more likely in lesions with high SUV values.Conclusion: The results of this study confirm that whole-body FDG-PET is more accurate than conventional CT in the staging of patients with suspected recurrent colorectal carcinoma.

Red marrow dosimetry for radiolabeled antibodies that bind to marrow, bone, or blood components.

Hematologic toxicity limits the radioactivity that may be administered for radiolabeled antibody therapy. This work examines approaches for obtaining biodistribution data and performing dosimetry when the administered antibody is known to bind to a cellular component of blood, bone, or marrow. Marrow dosimetry in this case is more difficult because the kinetics of antibody clearance from the blood cannot be related to the marrow. Several approaches for obtaining antibody kinetics in the marrow are examined and evaluated. The absorbed fractions and S factors that should be used in performing marrow dosimetry are also examined and the effect of including greater anatomical detail is considered. The radiobiology of the red marrow is briefly reviewed. Recommendations for performing marrow dosimetry when the antibody binds to the marrow are provided.

Use of PET to monitor the response of lung cancer to radiation treatment.

Approximately 170,000 people are diagnosed with lung cancer in the United States each year. Many of these patients receive external beam radiation for treatment. Fluorine-18 2-fluoro-2-deoxy-D-glucose positron emission tomography (FDG PET) is increasingly being used in evaluating non-small cell lung cancer and may be of clinical utility in assessing response to treatment. In this report, we present FDG PET images and data from two patients who were followed with a total of eight and seven serial FDG PET scans, respectively, through the entire course of their radiation therapy. Changes in several potential response parameters are shown versus time, including lesion volume (V(FDG)) by PET, SUVav, SUVmax, and total lesion glycolysis (TLG) during the course of radiotherapy. The response parameters for patient 1 demonstrated a progressive decrease; however, the response parameters for patient 2 showed an initial decrease followed by an increase. The data presented here may suggest that the outcome of radiation therapy can be predicted by PET imaging, but this observation requires a study of additional patients.

Tumor Burden Assessment with Positron Emission Tomography with

Objective: In patients with advanced cancer, total tumor burden affects the likelihood of tumor response and has important implications for prognosis. The aim of this study was to select the optimum 2-[F-18]fluoro-2-deoxy-D-glucose-positron emission tomography (FDG PET) tumor uptake parameter to accurately measure tumor burden in advanced metastatic renal cell cancer, in comparison with volumes measured with computed tomography (CT), as a reference test.Materials and Methods: Six patients with metastatic renal cell carcinoma measurable on CT were studied. CT and FDG PET scans were carried out on all patients within 4 weeks prior to their entry into a phase I-II radioimmunotherapy trial. CT-based evaluation of disease extent (tumor volume) and 4 PET-based measurements (standardized uptake value[SUVmax], SUVav, volume, and total lesion glycolysis [TLG]) were performed independently by a radiologist (VN) and a nuclear medicine physician (TA). The degree of correlation between conventional (CT) extent of disease and parameters describing tumor concentration of FDG was then determined.Results: Fifty-seven CT-measurable metastatic lesions in lung, abdomen, and scalp were evaluated in 6 patients. There was a high correlation between CT and FDG PET volume estimates for lesions greater than 5 cm(3) in size. However, a PET-derived parameter that embodies both FDG uptake and lesion size, the TLG, correlated better with CT-derived tumor volume than did FDG PET volume alone.Conclusion: Using CT volume as a gold standard, the optimal PET-based estimate of total tumor burden in patients with metastatic renal cancer is the sum over all lesions of the total lesion glycolysis.

Prospective assessment of primary rectal cancer response to preoperative radiation and chemotherapy using 18-fluorodeoxyglucose positron emission tomography.

PURPOSE: The purpose of this prospective study was to determine the ability of fluorine-18 fluorodeoxyglucose positron emission tomography to assess extent of pathologically confirmed rectal cancer response to preoperative radiation and 5-fluorouracil-based chemotherapy. METHODS: Patients with primary rectal cancer deemed eligible for preoperative radiation and 5-fluorouracil-based chemotherapy because of a clinically bulky or tethered tumor or endorectal ultrasound evidence of T3 and/or N1 were prospectively enrolled. Positron emission tomography and CT scans were obtained before preoperative radiation and 5-fluorouracil-based chemotherapy (5,040 cGy to the pelvis and 2 cycles of bolus 5-fluorouracil with leucovorin) and repeated four to five weeks after completion of radiation and 5-fluorouracil-based chemotherapy. In addition to routine pathologic staging, detailed assessment of rectal cancer response to preoperative radiation and 5-fluorouracil-based chemotherapy was performed independently by two pathologists. Positron emission tomography parameters studied included conventional measures such as standardized uptake value (average and maximum), positron emission tomography-derived tumor volume (size), and two novel parameters: visual response score and change in total lesion glycolysis. RESULTS: Of 21 patients enrolled, prospective data (pretreatment and posttreatment positron emission tomography, and complete pathologic assessment) were available on 15 patients. All 15 demonstrated pathologic response to preoperative radiation and 5-fluorouracil-based chemotherapy. This was confirmed in 100 percent of the cases by positron emission tomography compared with 78 percent (7/9) by CT. In addition, one positron emission tomography parameter (visual response score) accurately estimated the extent of pathologic response in 60 percent (9/15) of cases compared with 22 percent (2/9) of cases with CT. CONCLUSIONS: This pilot study demonstrates that fluorine-18 fluorodeoxyglucose positron emission tomography imaging adds incremental information to the preoperative assessment of patients with rectal cancer. However, further studies in a larger series of patients are needed to verify these findings and to determine the value of fluorine-18 fluorodeoxyglucose positron emission tomography in a preoperative strategy aimed at identifying patients suitable for sphincter-preserving rectal cancer surgery.

Pharmacokinetics and dosimetry of an alpha-particle emitter labeled antibody: 213Bi-HuM195 (anti-CD33) in patients with leukemia.

UNLABELLED: Data from nine patients with leukemia participating in a phase I activity-escalation study of HuM195, labeled with the alpha-particle emitter 213Bi (half-life = 45.6 min), were used to estimate pharmacokinetics and dosimetry. This is the first trial using an alpha-particle emitter in humans. The linear energy transfer of alpha particles is several hundredfold greater than that of beta emissions. The range in tissue is approximately 60-90 microm. METHODS: The activity administered to patients ranged from 0.6 to 1.6 GBq. Patient imaging was initiated at the start of each injection. Thirty 1-min images followed by ten 3-min images were collected in dynamic mode; a 20% photopeak window centered at 440 keV was used. Blood samples were collected until 3 h postinjection and counted in a gamma counter. Contours around the liver and spleen were drawn on the anterior and posterior views and around a portion of the spine on the posterior views. No other organs were visualized. RESULTS: The percentage injected dose in the liver and spleen volumes increased rapidly over the first 10-15 min to a constant value for the remaining hour of imaging, yielding a very rapid uptake followed by a plateau in the antibody uptake curves. The kinetic curves were integrated to yield cumulated activity. The mean energy emitted per nuclear transition for 213Bi and its daughters, adjusted by a relative biologic effectiveness of 5 for alpha emissions, was multiplied by the cumulated activity to yield the absorbed dose equivalent. Photon dose to the total body was determined by calculating a photon-absorbed fraction. The absorbed dose equivalent to liver and spleen volumes ranged from 2.4 to 11.2 and 2.9 to 21.9 Sv, respectively. Marrow (or leukemia) mean dose ranged from 6.6 to 12.2 Sv. The total-body dose (photons only) ranged from 2.2 x 10(-4) to 5.8 x 10(-4) Gy. CONCLUSION: This study shows that patient imaging of 213Bi, an alpha-particle emitter, labeled to HuM195 is possible and may be used to derive pharmacokinetics and dosimetry. The absorbed dose ratio between marrow, liver and spleen volumes and the whole body for 213Bi-HuM195 is 1000-fold greater than that commonly observed with beta-emitting radionuclides used for radioimmunotherapy.

Use of the fast Hartley transform for three-dimensional dose calculation in radionuclide therapy.

Effective radioimmunotherapy may depend on a priori knowledge of the radiation absorbed dose distribution obtained by trace imaging activities administered to a patient before treatment. A new, fast, and effective treatment planning approach is developed to deal with a heterogeneous activity distribution. Calculation of the three-dimensional absorbed dose distribution requires convolution of a cumulated activity distribution matrix with a point-source kernel; both are represented by large matrices (64 x 64 x 64). To reduce the computation time required for these calculations, an implementation of convolution using three-dimensional (3-D) fast Hartley transform (FHT) is realized. Using the 3-D FHT convolution, absorbed dose calculation time was reduced over 1000 times. With this system, fast and accurate absorbed dose calculations are possible in radioimmunotherapy. This approach was validated in simple geometries and then was used to calculate the absorbed dose distribution for a patient's tumor and a bone marrow sample.

A new parameter for measuring metastatic bone involvement by prostate cancer: the Bone Scan Index.

In this report, we describe a method for quantitative bone scan interpretation (the Bone Scan Index or BSI) in advanced prostate cancer. The BSI estimates the fraction of the skeleton that is involved by tumor, as well as the regional distribution of the metastases in the bones. The purpose of this report is to describe the development and validation of this method in terms of reproducibility and the application of BSI for determining extent of disease and monitoring disease progression. We analyzed 263 bone scans from 90 patients being studied under four protocols at Memorial Sloan-Kettering Cancer Center for progressive, androgen-independent prostate cancer (AIPC), who had bone scans as a part of their work-up. We determined: (a) the intraobserver and interobserver variability of the BSI; (b) the comparison between a change in BSI and prostate-specific antigen (PSA); (c) the regional distribution of bony metastases in early stage D prostate cancer (<3% skeletal involvement); and (d) the rate of growth of bony metastases from prostate cancer. A cube root transformation of the percentage of involvement of the entire skeleton was used to stabilize the variance over the entire span of values (0-60% tumor involvement). The range of interobserver variability between readers was 0.2-0.5 times the cube root of the BSI (69 scans, 18 patients). Intraobserver variability was minimal when the same reader read the same scans after a 2-year interval, showing a correlation coefficient of 0.97 (reader 1) and 0.99 (reader 2), P < 0.001. There was a parallel rise in the BSI and the PSA in 24 patients (105 scans) treated for AIPC with hydrocortisone followed by suramin at PSA relapse (Pearson's moment correlation, 0.71). In a group of 27 patients with limited bone involvement by AIPC (i.e., <3% BSI), the distribution of early metastases was not random within the skeleton but was distributed in the central skeleton in a manner that matched the distribution of the normal adult bone marrow. Also, in a group of 21 patients (62 scans), the change in BSI as a function of time after diagnosis was explored graphically. The progression of bone scan changes in AIPC, from early involvement (<3%) to late involvement, was fitted to a Gompertzian equation. It showed a rapid exponential growth phase, with an estimated tumor doubling time of 43 days when the BSI was 3.3%. The change in BSI rapidly approached a more gradual slope as the percentage of skeletal involvement increased. The BSI provides a reproducible new parameter for quantitative assessment of bone involvement by AIPC. These results suggest that the BSI will be useful for stratifying patients entering treatment protocols for extent of tumor involvement of bone. Although further study is necessary, serial bone scan BSI appears capable of quantifying both the progression of bony involvement by tumor as well as the response to treatment.

Segmentation of lung lesion volume by adaptive positron emission tomography image thresholding.

BACKGROUND: It is common protocol in radionuclide therapies to administer a tracer dose of a radiopharmaceutical, determine its lesion uptake and biodistribution by gamma imaging, and then use this information to determine the most effective therapeutic dose. This treatment planning approach can be used to quantitate accurately the activity and volume of lesions and organs with positron emission tomography (PET). In this article, the authors focus on the specification of appropriate volumes of interest (VoI) using PET in association with computed tomography (CT). METHODS: The authors have developed an automatic image segmentation schema to determine the VoI of metastases to the lung from PET images, under conditions of variable background activity. An elliptical Jaszczak phantom containing a set of spheres with volumes ranging from 0.4 to 5.5 mL was filled with F-18 activity (2-3 microCi/mL) corresponding to activities clinically observed in lung lesions. Images were acquired with a cold background and then with variable source-to-background (S/B) ratios of: 7.4, 5.5, 3.1, and 2.8. Lesion VoI analysis was performed on 10 patients with 17 primary or metastatic lung lesions, applying the optimum threshold values derived from the phantom experiments. Initial volume estimates for lung lesions were determined from CT images. Approximate S/B ratios were obtained for the corresponding lesions on F-18-fluoro-2-deoxy-D-glucose (18FDG)-PET images. From the CT estimate of the lesion size and the PET estimate of the S/B ratio, the appropriate optimum threshold could be chosen. The threshold was applied to the PET images to obtain lesion activity and a final estimate of the lesion volume. RESULTS: Phantom data analysis showed that image segmentation converged to a fixed threshold value (from 36% to 44%) for sphere volumes larger than 4 mL, with the exact value depending on the S/B ratios. For patients, the use of optimum threshold schema demonstrated a good correlation (r = 0.999) between the initial volume from CT and the final volume derived from the 18FDG-PET scan (P < 0.02). The mean difference for those volumes was 8.4%. CONCLUSIONS: The adaptive thresholding method applied to PET scans enables the definition of tumor VoI, which hopefully leads to accurate tumor dosimetry. This method can also be applied to small lesions (<4 mL). It should enable physicians to track objectively changes in disease status that could otherwise be obscured by the uncertainties in the region-of-interest drawing, even when the scans are delineated by the same physician.

Quantitative bone metastases analysis based on image segmentation.

UNLABELLED: Preliminary evidence indicates that the fraction of bone containing metastatic lesions is a strong prognostic indicator of survival longevity for prostate and breast cancer. Our current approach to quantify metastatic bone lesions, called the Bone Scan Index, is based on an inspection of the bone scan, estimating visually the fraction of each bone involved and then summing across all bones to determine the percentage of total skeletal involvement. This approach, however, is time consuming, subjective and dependent on individual interpretation. METHODS: To overcome these problems, a semiautomated image segmentation program was developed for the quantitation of metastases from planar whole-body bone scans. The user is required to insert a seed point into each metastatic region on the image. The algorithm then connects pixels to the seed pixel in all directions until a contrast-dependent threshold is reached. The optimal threshold for cessation of the region growing is determined from phantom studies. On the images, lesion delineation and size measurements were performed by the algorithm. Each delineated lesion is associated with a bone site using pull-down menus. The program then computes the fraction of lesion involvement in each bone based on look-up-tables containing the relationship of bone mass with race, sex, height and age. These look-up-tables were obtained by multiple regression of the skeletal mass measurements in humans. The total fraction of skeletal involvement is then obtained from the individual fractional masses. For individual fractional mass, values given in International Commission on Radiation Protection Publication No. 23 were used. RESULTS: The bone metastases analysis system has been used on 11 scans from 6 patients. The correlation was high (r = 0.83) between conventional (manually drawn region-of-interest) and this analysis system. Bone metastases analysis results in consistently lower estimates of fractional involvement in bone compared with the conventional region-of-interest drawing or visual estimation method. This is due to the apparent broadening of objects at and below the limits of resolution of the gamma camera. CONCLUSION: Image segmentation reduces the delineation and quantitation time of lesions by at least two compared with manual region-of-interest drawing. The objectivity of this technique allows the detection of small variations in follow-up patient scans for which the manual region-of-interest method may fail, due to performance variability of the user. This method preserves the diagnostic skills of the nuclear medicine physician to select which bony structures contain lesions, yet combines it with an objective delineation of the lesion.