Evaluation of image quality, radiation dose, and tumor size assessment in chest tomosynthesis

<jats:p>&lt;p dir="ltr"&gt;Digital tomosynthesis (DTS) is an X-ray technique that provides sectional images with better lesion visibility than conventional chest radiology (CXR), while using for less radiation than CT. Optimized DTS protocols can maintain high thoracic image quality, including vessels, bone structures and tumor morphology, even at reduced dose. Together with low-cost and easy integrations into standard radiography equipment, this makes DTS a promising alternative for detection and follow-up of pulmonary consolidations and lung malignancies. The purpose of this thesis was to evaluate digital chest tomosynthesis (DTS) as a low-dose imaging technique in thoracic radiology by comparing image quality across acquisition and post-processing protocols, assessing the possibilities for radiation dose reduction. A further aim was to determine the accuracy and reliability of lung consolidation size measurements relative to computed tomography (CT).&lt;/p&gt;&lt;p dir="ltr"&gt;Paper I- A chest phantom was imaged on a new chest DTS system using ten different acquisition protocols at the same effective dose, with nine acquisitions per protocol. Four observers visually graded the reconstructed images using predefined quality criteria in four classes: demarcation, disturbance, structure, and homogeneity in nodule. Image quality was analyzed with visual grading characteristics (VGC), using the vendor recommended protocol (12 s) as the reference and the area under the VGC curve (AUGvGc) as the figure-of- merit. The 6.3 s protocol yielded image quality comparable to the 12 s vendor recommended protocol and offered the clinically important benefit of a shorter acquisition time for patients with respiratory limitations. This indicates that the vendor recommended protocol may not be the optimal choice. This study showed that a shorter protocol can be a viable alternative to the vendor recommended protocol, providing the basis for the clinical evaluation in paper II.&lt;/p&gt;&lt;p dir="ltr"&gt;Paper II- Twenty patients with suspected lung cancer were examined with chest DTS using two protocols with different acquisitions time (6.3 s and 12 s). Each protocol was reconstructed with two post-processing options (standard DTS and advance post-processing for chest radiography), resulting in four anonymized, randomly ordered image series per patient. Five observers rated image quality according to predefined criteria in three classes (demarcation, disturbance and structure), and VGC analysis was performed with 12 s protocol and the standard DTS post-processing as references. The 6.3s protocol showed a statistically significant image quality advantage over the 12s protocol for demarcation (AUCVGC = 0.56, p = 0.009) and disturbance (AUGvGc = 0.58, p&lt; 0.001), while for structure (bone definition in the spine) the AUGvac was 0.56 but did not differ statistically significant from 0.5 (p = 0.21). Standard DTS post- processing protocol had a small but statistically significant advantage over the advanced processing for demarcation (AUGvGc = 0.45, p = 0.017) and disturbance (AUGvGc = 0.43, P = 0.005), whereas a similar value for structure (0.46) was not significantly different from 0.5 (p = 0.31). Overall, a protocol with 6.3s acquisition time provides slightly better image quality than the vendor recommended 12s protocol for several anatomical structures, and the standard processing yields some advantage over the more advanced post processing. This reinforces that DTS protocol and post-processing choices have a clear impact on image quality, and the vendor recommended settings are not always optimal.&lt;/p&gt;&lt;p dir="ltr"&gt;These findings motivated the subsequent design of papers III and IV. In paper III, the focus was to investigate whether low-dose DTS can be used for follow-up examinations without compromising clinically important image quality, and to explore its suitability in lung cancer screening settings. In paper IV, the accuracy of lung consolidation measurements with low-dose DTS was evaluated in comparison with CT within the same clinical context, and to explore DTS as a more accessible lower dose alternative to CT for follow-up and potentially, for use in lung cancer screening programs.&lt;/p&gt;&lt;p dir="ltr"&gt;Paper III- Fifty patients with known or suspected lung malignancy were imaged with DTS using three protocols, the vendor recommended dose and two low- dose protocols with 30% and 50% dose reduction. Four blinded observers rated the quality of reconstructed coronal section images according to predefined criteria in several thoracic regions (vessel demarcation, disturbance of vessels, bone structure), and tumor homogeneity. VGC analysis with AUCvGc as figure-of- merit was applied. Compared with the vendor recommended protocol, the 30% low-dose protocol showed no reduction in image quality for any class of criteria (demarcation AUCvGc = 0.51, p = 0.57; disturbance AUCvGc = 0.51, p = 0.59; bone structure AUCVGC = 0.50, p = 0.88 and tumor homogeneity AUCvGc = 0.47, p = 0.09), whereas the 50% low-dose protocol yielded reduced image quality for demarcation (AUCVGC = 0.47, p = 0.03) and disturbance (AUCVGc = 0.47, p = 0.02), with no significant degradation for bone structure (AUCvGc = 0.50, p = 0.79) or tumor homogeneity ( AUCvGc = 0.46, p = 0.10). Overall, these results indicate that reducing the DTS dose by 30% does not reduce clinically relevant image quality and appears feasible in clinical practice, while a 50% reduction introduces modest but statistically significant deterioration in certain vessel-related criteria, although bone and tumor visualization remains acceptable.&lt;/p&gt;&lt;p dir="ltr"&gt;Paper IV- This retrospective study included 49 patients with at least one lung consolidation, among the 50 initially included patients, one of whom was excluded due to technical error. Clinically indicated CT was used as reference, and no additional CT examinations were performed for research purpose. Each patient underwent chest DTS with a standard dose protocol (0.13 mSv) and a protocol with 30% dose reduction (0.09 mSv). Between one and three clearly visible consolidations per patient were selected on CT and measured in horizontal and craniocaudal dimensions. Four radiologists without prior DTS experience, blinded to CT and other information, independently measured the same consolidations on randomized, anonymized DTS series. Mean DTS measurements across observers were compared with CT using t-tests, and inter-observer reliability was assessed with intraclass correlation coefficients (ICC). Agreement between standard-dose (Tomo1) and low-dose (Tomo2) DTS was high, with very similar mean consolidation size and only a small, statistically significant but clinically minor advantage for Tomo1 in the craniocaudal dimension. In addition, both DTS protocols showed very good agreement with CT, with mean size estimates close to CT, small absolute and relative differences, and DTS-to-CT ratio mean 1.0 across size subgroups. Inter-observer reliability for DTS measurements was excellent, with single-measure ICCs above 0.90 and average measure ICCs approaching 0.98, indication that most variability reflected true size differences rather than reader disagreement. Overall, the study showed that low-dose chest DTS provides images that allow pulmonary consolidation size to be assessed with accuracy and consistency comparable to standard-dose DTS and CT. This supports its use as a robust, lower-dose and more accessible alternative to CT for follow-up and for possible use in lung cancer screening programs.&lt;/p&gt;&lt;h3 dir="ltr"&gt;List of scientific papers&lt;/h3&gt;&lt;p dir="ltr"&gt;I. &lt;b&gt;Jadidi M,&lt;/b&gt; Sundin A, Aspelin P, Båth M, Nyrén S. Evaluation of a new system for chest tomosynthesis: aspects of image quality of different protocols determined using an anthropomorphic phantom. Br J Radiol. 2015;88(1053):20150057. Included in the previous licentiate thesis. &lt;a href="https://doi.org/10.1259/bjr.20150057" target="_blank" rel="noreferrer"&gt;https://doi.org/10.1259/bjr.20150057&lt;/a&gt;&lt;/p&gt;&lt;p dir="ltr"&gt;II. &lt;b&gt;Jadidi M,&lt;/b&gt; Bath M, Nyren S. Dependency of image quality on acquisition protocol and image processing in chest tomosynthesis-a visual grading study based on clinical data. Br J Radiol. 2018; 91(1087):20170683. Included in the previous licentiate thesis. &lt;a href="https://doi.org/10.1259/bjr.20170683" target="_blank" rel="noreferrer"&gt;https://doi.org/10.1259/bjr.20170683&lt;/a&gt;&lt;/p&gt;&lt;p dir="ltr"&gt;III. &lt;b&gt;Jadidi M,&lt;/b&gt; Båth M, Svalkvist, A, Nyrén S. Can the current radiation dose, in chest tomosynthesis, be reduced with retained image quality? A study in the context of lung cancer screening programs. PloS One 2026;21(3).&lt;br&gt;&lt;a href="https://doi.org/10.1371/journal.pone.0343760"&gt;https://doi.org/10.1371/journal.pone.0343760&lt;br&gt;&lt;/a&gt;&lt;br&gt;&lt;/p&gt;&lt;p dir="ltr"&gt;IV. &lt;b&gt;Jadidi M,&lt;/b&gt; Båth M, Svalkvist, A, Nyrén S. Comparative evaluation of pulmonary consolidation size measurements on standard- and low-dose chest tomosynthesis using CT as the reference standard. [Submitted]&lt;/p&gt;</jats:p>