Automatic lung segmentation in COVID-19 patients: Impact on quantitative computed tomography analysis.

PURPOSE: To assess the impact of lung segmentation accuracy in an automatic pipeline for quantitative analysis of CT images. METHODS: Four different platforms for automatic lung segmentation based on convolutional neural network (CNN), region-growing technique and atlas-based algorithm were considered. The platforms were tested using CT images of 55 COVID-19 patients with severe lung impairment. Four radiologists assessed the segmentations using a 5-point qualitative score (QS). For each CT series, a manually revised reference segmentation (RS) was obtained. Histogram-based quantitative metrics (QM) were calculated from CT histogram using lung segmentationsfrom all platforms and RS. Dice index (DI) and differences of QMs (ΔQMs) were calculated between RS and other segmentations. RESULTS: Highest QS and lower ΔQMs values were associated to the CNN algorithm. However, only 45% CNN segmentations were judged to need no or only minimal corrections, and in only 17 cases (31%), automatic segmentations provided RS without manual corrections. Median values of the DI for the four algorithms ranged from 0.993 to 0.904. Significant differences for all QMs calculated between automatic segmentations and RS were found both when data were pooled together and stratified according to QS, indicating a relationship between qualitative and quantitative measurements. The most unstable QM was the histogram 90th percentile, with median ΔQMs values ranging from 10HU and 158HU between different algorithms. CONCLUSIONS: None of tested algorithms provided fully reliable segmentation. Segmentation accuracy impacts differently on different quantitative metrics, and each of them should be individually evaluated according to the purpose of subsequent analyses.

A patient-specific approach for quantitative and automatic analysis of computed tomography images in lung disease: Application to COVID-19 patients.

PURPOSE: Quantitative metrics in lung computed tomography (CT) images have been widely used, often without a clear connection with physiology. This work proposes a patient-independent model for the estimation of well-aerated volume of lungs in CT images (WAVE). METHODS: A Gaussian fit, with mean (Mu.f) and width (Sigma.f) values, was applied to the lower CT histogram data points of the lung to provide the estimation of the well-aerated lung volume (WAVE.f). Independence from CT reconstruction parameters and respiratory cycle was analysed using healthy lung CT images and 4DCT acquisitions. The Gaussian metrics and first order radiomic features calculated for a third cohort of COVID-19 patients were compared with those relative to healthy lungs. Each lung was further segmented in 24 subregions and a new biomarker derived from Gaussian fit parameter Mu.f was proposed to represent the local density changes. RESULTS: WAVE.f resulted independent from the respiratory motion in 80% of the cases. Differences of 1%, 2% and up to 14% resulted comparing a moderate iterative strength and FBP algorithm, 1 and 3 mm of slice thickness and different reconstruction kernel. Healthy subjects were significantly different from COVID-19 patients for all the metrics calculated. Graphical representation of the local biomarker provides spatial and quantitative information in a single 2D picture. CONCLUSIONS: Unlike other metrics based on fixed histogram thresholds, this model is able to consider the inter- and intra-subject variability. In addition, it defines a local biomarker to quantify the severity of the disease, independently of the observer.

Performance comparison of two resolution modeling PET reconstruction algorithms in terms of physical figures of merit used in quantitative imaging.

PURPOSE: Resolution modeling (RM) of PET systems has been introduced in iterative reconstruction algorithms for oncologic PET. The RM recovers the loss of resolution and reduces the associated partial volume effect. While these methods improved the observer performance, particularly in the detection of small and faint lesions, their impact on quantification accuracy still requires thorough investigation. The aim of this study was to characterize the performances of the RM algorithms under controlled conditions simulating a typical (18)F-FDG oncologic study, using an anthropomorphic phantom and selected physical figures of merit, used for image quantification. METHODS: Measurements were performed on Biograph HiREZ (B_HiREZ) and Discovery 710 (D_710) PET/CT scanners and reconstructions were performed using the standard iterative reconstructions and the RM algorithms associated to each scanner: TrueX and SharpIR, respectively. RESULTS: RM determined a significant improvement in contrast recovery for small targets (≤17 mm diameter) only for the D_710 scanner. The maximum standardized uptake value (SUVmax) increased when RM was applied using both scanners. The SUVmax of small targets was on average lower with the B_HiREZ than with the D_710. Sharp IR improved the accuracy of SUVmax determination, whilst TrueX showed an overestimation of SUVmax for sphere dimensions greater than 22 mm. The goodness of fit of adaptive threshold algorithms worsened significantly when RM algorithms were employed for both scanners. CONCLUSIONS: Differences in general quantitative performance were observed for the PET scanners analyzed. Segmentation of PET images using adaptive threshold algorithms should not be undertaken in conjunction with RM reconstructions.

An analytical model for improving absorbed dose calculation accuracy in non spherical autonomous functioning thyroid nodule.

AIM: Patients candidate to radioiodine treatment of autonomous functioning thyroid nodule (AFTN) are characterized by a wide range of nodule volumes with different shapes. To optimize the treatment, pretherapeutic dosimetry should account also for the dependence of deposited energy on the nodule geometry. METHODS: We developed a Monte Carlo code in Geant4 to simulate the interaction of beta and gamma radiations emitted by Na-131I into ellipsoidal volumes of soft tissue homogeneously uptaking the radionuclide, surrounded by a simplified antropomorphic phantom. We simulated 9 volumes between 0.1 and 50 cm3, each one with 8 different ellipsoidal shapes. We considered the data of 10 patients affected by AFTN, whose nodule volumes were in the range 1-40 cm3, who underwent radioiodine therapy following the traditional dosimetric approach. The patients underwent ultrasonographic (US) study, in order to determine the nodule volume, and radioiodine thyroid uptake measurements between 3 and 168 hours after radioiodine tracer dose administration. RESULTS: We found an analytical relationship between the average deposited energy and the ellipsoid's semiaxes and we included it in the formula for the calculation of activity to be administered, A0. For the 10 patients studied, A0 calculated with our approach ranges from +9% to -2% with respect to the one calculated with the traditional formula. CONCLUSION: The proposed model, accounting for the dependence of beta and gamma absorbed fractions from nodule volume and shape, can lead to a more accurate estimation of A0 during AFTN therapy. Since the measurement of nodule axes is routinely obtained from pretherapeutic US, our approach can be introduced in the clinical practice without changing the diagnostic pre-therapeutic protocol.

Absorbed fractions for electrons in ellipsoidal volumes.

We applied a Monte Carlo simulation in Geant4 in order to calculate the absorbed fractions for monoenergetic electrons in the energy interval between 10 keV and 2 MeV, uniformly distributed in ellipsoids made from soft tissue. For each volume, we simulated a spherical shape, four oblate and four prolate ellipsoids, and one scalene shape. For each energy and for every geometrical configuration, an analytical relationship between the absorbed fraction and a 'generalized radius' was found, and the dependence of the fit parameters from electron energy is discussed and fitted by proper parametric functions. With the proposed formulation, the absorbed fraction for electrons in the 10-2000 keV energy range can be calculated for all volumes and for every ellipsoidal shape of practical interest. This method can be directly applied to evaluation of the absorbed fraction from the radionuclide emission of monoenergetic electrons, such as Auger or conversion electrons. The average deposited energy per disintegration in the case of extended beta spectra can be evaluated through integration. Two examples of application to a pure beta emitter such as (90)Y and to (131)I, whose emission include monoenergetic and beta electrons plus gamma photons, are presented. This approach represent a generalization of our previous studies, allowing a comprehensive treatment of absorbed fractions from electron and photon sources uniformly distributed in ellipsoidal volumes of any ellipticity and volume, in the whole range of practical interest for internal dosimetry in nuclear medicine applications, as well as in radiological protection estimations of doses from an internal contamination.

Absorbed fractions for photons in ellipsoidal volumes.

We studied through Monte Carlo simulation in Geant4 the absorbed fractions for photons, characterized by energies ranging from 10 keV to 1000 keV, which can be emitted by gamma radionuclides uniformly distributed in ellipsoidal volumes of soft tissue. The same analytical relationship between absorbed fraction and the 'generalized radius' as introduced in a previous paper was found, and the dependence of its parameters rho(0) and s on photon energy is discussed and fitted by suitably chosen parametric functions. As a consequence, the absorbed fraction for photons in the 10-1000 keV energy range can be calculated for all volumes and for every ellipsoidal shape of practical interest. Such results can be a useful complement for the dosimetry of beta- and gamma-emitting radionuclides during internal radiotherapy or gamma emitters employed in diagnostic nuclear medicine.

Absorbed fractions in ellipsoidal volumes for beta(-) radionuclides employed in internal radiotherapy.

We developed a Monte Carlo simulation in Geant4 to calculate the absorbed fractions for electrons emitted by (199)Au, (177)Lu, (131)I, (153)Sm, (186)Re and (90)Y, characterized by average energies ranging from 86 keV to 949 keV, uniformly distributed in ellipsoidal volumes of soft tissue. Code validation results with respect to reference data for doses, ranges and absorbed fractions in spheres are presented. An analytical relationship between the absorbed fraction and a 'generalized radius' is introduced in analogy with the transfer function of a first-order high-pass filter, and the dependence of its parameters rho(0) and s from the average electron energy and range is discussed. A generalization for the estimation of absorbed fractions for other radionuclides is also proposed. Such results can be useful to improve accuracy and easiness of calculation in dosimetry during internal radiotherapy.

Plastic materials as a radiation shield for beta- sources: a comparative study through Monte Carlo calculation.

We have developed a Monte Carlo simulation in Geant4 to compare the attenuation properties and the bremsstrahlung radiation yield of different types of plastic materials employed as shields for beta- radioactive sources. Code validation results against Sandia and NIST data are presented. For polypropylene (C3H6), polystyrene (C2H3), polyamide nylon-6 (C6H11ON), poly-methyl methacrylate (C5H8O2), polycarbonate (C16H6O3), polyethylene terephthalate (C10H8O4), polyvinyl chloride (C2H3Cl) and polytetrafluoroethylene (C2F4) we evaluated the mean and maximum ranges for electrons originating from 90Sr and 90Y, as well as the number and spectrum of the bremsstrahlung x-rays produced. Significant differences appear between the various materials, and the choice of the best one also depends on the physical properties requested for each specific application.