Digital Tomosynthesis as a Problem-Solving Technique to Confirm or Exclude Pulmonary Lesions in Hidden Areas of the Chest

Abstract

Objectives: To evaluate the capability of digital tomosynthesis (DTS) to characterize suspected pulmonary lesions in the so-called hidden areas at chest X-ray (CXR).

Materials and methods: Among 726 patients with suspected pulmonary lesions at CXR who underwent DTS, 353 patients (201 males, 152 females; age 71.5 ± 10.4 years) revealed suspected pulmonary lesions in the apical, hilar, retrocardiac, or paradiaphragmatic lung zones and were retrospectively included. Two readers analyzed CXR and DTS images and provided a confidence score: 1 or 2 = definitely or probably benign pulmonary or extra-pulmonary lesion, or pulmonary pseudo-lesion deserving no further diagnostic work-up; 3 = indeterminate lesion; 4 or 5 = probably or definitely pulmonary lesion deserving further diagnostic work-up by CT. The nature of DTS findings was proven by CT (n = 108) or CXR during follow-up (n = 245).

Results: In 62/353 patients the suspected lung lesions were located in the lung apex, in 92/353 in the hilar region, in 59/353 in the retrocardiac region, and in 140/353 in the paradiaphragmatic region. DTS correctly characterized the CXR findings as benign pulmonary or extrapulmonary lesion (score 1 or 2) in 43/62 patients (69%) in the lung apex region, in 56/92 (61%) in the pulmonary hilar region, in 40/59 (67%) in the retrocardiac region, and in 106/140 (76%) in the paradiaphragmatic region, while correctly recommending CT in the remaining cases due to the presence of true solid pulmonary lesion, with the exception of 22 false negative findings (60 false positive findings). DTS showed a significantly (p < 0.05) increased sensitivity, specificity, and overall diagnostic accuracy and area under ROC curve compared to CXR alone.

Conclusions: DTS allowed confirmation or exclusion of the presence of true pulmonary lesions in the hidden areas of the chest.

Keywords: chest digital tomosynthesis; pulmonary hidden areas; pulmonary lesions.

Figure 1

Flow diagram of the study population.

Figure 2

Pre-operative X-ray for colorectal carcinoma of a 56-year-old man. (A) Posteroanterior chest radiography in the upright position shows one suspected pulmonary nodule in the right lung base (arrow). (B–D) Digital tomosynthesis images confirm the previous nodule and identifies additional nodules (arrows) in both lungs, subsequently confirmed by CT (E–G).

Figure 3

Pre-operative X-ray of a 72-year-old woman for skin melanoma. (A) Posteroanterior chest radiography in the upright position shows one suspected pulmonary nodule in the left apex (arrow). (B) Digital tomosynthesis image clarifies that the same opacity corresponds to a costal arthrosic hypertrophy of the anterior arch of the first left rib (arrow). Readers provided a confidence score of 1.