Table 5.
Main region-of-interest specifications
| First author | 2D/3D | Nr. (one/multiple)1 | Predefined absolute size? | Excluding areas2 | Only highest signal (b-image)/lowest on ADC map/solid part tumor3 | ROI no residual disease post-NST visible |
|---|---|---|---|---|---|---|
| Santamaria [27]4 | 2D: circular | Three (diff. sections) | Y: (≤ 15 mm2) | Y | Not specified | Pretreatment location |
| Tozaki [40] | 2D: circular | One | Y 19.6 mm2 (r = 2.5 mm) | n.r. | Y | n.r. |
| Bufi [17] | ||||||
| – Before | 2D | One | No | Y | No | N/A |
| – Post5 | – | – | – | – | – | |
| Che [19] | 2D | One | No: based on max transverse diameter | Y | Different description | n.r. |
| Fangberget [65] | Not spec. | One (solid part) | No | Y | Y | n.r. |
| Minarikova [59]6 | 3D | One | No: region growing (upper & lower bounds) | Not specified (in 3D) | No | n.r. |
| Woodhams [64]7 | 2D | Pre: two to seven | No | Not specified | No | n.r. |
| Post: one to seven | ||||||
| Shin [26] | 3D | One | No | Y | No | No residual enhancement: images compared pre and post-NAC, incl. cardiac level and the surrounding |
| Hahn [67] | 2D | Three (slices) | No (based on the largest cross-sectional planes → three slides) | Y (especially fat and normal parenchyma, further not specified) | No | n.r. |
| Yuan [22] | 3D | One | No | Y | No | n.r. |
| Partridge [23] | 3D | One composite | No | Y | No | Region at previous scan with visible tumor |
| Gallivanone [21] | 3D | One | No (semi-automatic method: see Gallivanone et al.) | Y | Not only (see details Gallivanone et al.) | n.r. |
| Li [44] | 3D | One | No (copied from the DCE-ROI tumor8) | n.r. | Different description | n.r. |
| Fujimoto [66] | 2D | One | No (based on largest diameter) | Y | Different description | Pre-treatment ROI |
| Liu [16] | 2D (pseudo- 3D) | Three | No (based on largest cross-sectional area’s) | n.r. | Different description | Pre-treatment ROI |
| Bedair [20] | 2D | One | No | Y | No: largest tumor dimension on b = 900 | n.r. |
| Ramirez-Galván [25] | 2D | Three | No (three ellipses randomly placed) | Y | No | ? |
| Pereira [18] | 2D | One | n.r. | Y | Y | n.r. |
| Zhang [24] | 2D | Three types: | n.r. | Y | Different description | n.r. |
| (a) Freehand | ||||||
| (b) Single-round | ||||||
| (c) Three-round | ||||||
| Kim [53] | 2D (manual) → 3D (automatic) | Three (sagittal, coronal, axial) | No | Y | Different (on b = 0, based on DCE and T2) | n.r. |
ADC apparent diffusion coefficient, n.r. not reported, r radius, ROI region of interest
1Short description
2Not specified for which areas (but referring to areas such as inner margins, necrotic, fibrotic areas etc.)
3No: the ROI was not limited to solid or other tumor part on the slice or high signal on b-image, respectively, low signal on ADC
4Multiple ROI methods, but this refers to mean value method used for data analysis, the reported area could be for all three together or each area
5Situation depended: delineation on b = 1000 s/mm2, in case of tumor fragmentation ROI including not hyperintense “interspersed” area and the whole lesion, otherwise in case of no clear region, ROI of 100 pixels within the previous observed area
62D and 3D ROI’s: 3D-ROI’s in majority used for comparisons-not applicable: referring to a different image property
7The ADC of the multiple ROI’s were averaged and mean ROI-size was pre-NST 37 ± 17 mm and post-NST 20 ± 15 mm
8At least 80% percent signal intensity increase after contrast injection, see further Li et al. [44] Pseudo 3D: several slices, but not whole lesion in 3D