Axial ROI: Using the reference implant location line, the ROI was defined as the alveolar bone area with a 6-mm width corresponding to 3 mm on each side of the future implant center, including the buccal and lingual cortical layers.

Coronal ROI: Area defined by the outer border of the cortical bone with a 6-mm height and a line joining the buccal and lingual cortical layers.

Sagittal ROI: A 6 × 6 mm square was defined as the area corresponding to the implant plus 1 mm of the surrounding bone at the mesial and distal sides.

The visual evaluation of bone quality in the preoperative CT images was performed by a senior, calibrated, and experienced radiologist in CT, who was blind to the other measures of bone quality and insertion torque. After memorizing the ROIs traced in the axial, coronal, and sagittal CT sections, the boundary lines were removed to avoid any interference during the visual assessment. The cortical bone was defined as a white and homogeneous outer layer of the alveolar ridge. The trabecular bone was defined as the structure between the two cortical layers. The examiner analyzed the images as many times as needed to categorize each implant site into bone types 1, 2, 3, or 4, according to the classification of Lekholm and Zarb [5].

The DICOM files were imported to the ImageJ software, and the original axial stacks were used for CT image normalization in 32-bit [18]. Two 5 × 5 mm squares were delimitated for air and soft tissue in the same axial slice containing the ROI (cortical plus trabecular bones).

The original CT scans were rescaled applying the following formula:

where −GV_air is the mean gray value for air and GV_st is the mean gray value for a central soft tissue square. This calibration was performed by subtracting the gray value for air and multiplying the result by the ratio of 1000 divided by the result of the gray value for soft tissue minus the gray value for air. As a result of this gray value transformation, GV_air = 0 and GV_st = 1000 for all CT images.