3T MRI of the breast with computer aided diagnosis, can it help to avoid unnecessary invasive procedures???

Peer review under responsibility of Alexandria University Faculty of Medicine.

Available online 14 January 2015

Figures & data

Table 1 Scoring points and sum for biopsy proven (BP) benign and malignant lesions.²¹

Character	Score	Average for BP. benign	Average for BP. malignant
T2
Hyperintense	−3	−2.2	0.2
Hypointense	1
Edema
No edema	−1	−0.8	1.3
Bilateral, bifocal, diffuse	1
Unifocal, unilateral	2
Hook sign
Positive	1	0	0.3
Negative	0
Skin thickening
Positive	1	0.1	0.6
Negative	0
DWI
Restricted diffusion	2	0.6	1.6
No restriction	0
Vascular score
Absent	0
Low	1	1.8	1.3
Moderate	2
High	3
Margins of contrast enhanced lesion
Well-defined	0	0.3	0.6
Ill-defined	1
Shape of contrast-enhanced lesion
Round, lobular, oval	0	0	0.7
Linear, dendritic, stellate	1
Enhancement pattern
Homogeneous	0	0.1	0.8
Inhomogeneous	1
Initial peak signal intensity increase
<50%	0		1.8
50–100%	1
>100%	2	0.5
DCE curve
Continuous raise	0	0.3	1.7
Plateau	1
Wash-out	2
Lymph nodes	0
No lymph nodes	0	0.6	0.7
<10 mm	1
>10 mm
Sum of average scoring points	1.3		11.6

Figure 1 (A) Axial T2w image showing left breast hypointense (proven malignant) nodule (arrow) and right breast incidentally discovered small well defined benign looking hyperintense nodules (arrowheads). (B) Another patient with right breast malignant T2w hypointense infiltrative tumor with dendritic outline (arrow).

Figure 2 (A and B). Left breast proven malignant tumor: axial STIR images show left breast long TR hypointense swelling (arrowhead) with localized perifocal edema, asymmetrical skin thickening (arrow in B), dendritic infiltrative outline and positive hook sign (curved arrow).

Figure 3 (A) T2w image, of right breast proven malignant tumor, shows irregular hypointense lesion (arrow), (B) ADC the lesion shows remarkable restricted diffusion, seen as low signal intensity. (C) Another patient with left breast benign swellings showing T2w Hyperintense signal (arrows). (D) No restricted diffusion at all in DWI.

Figure 4 (A) US left breast showing central hypoechoic soft tissue lesion with highly suspicious malignant features (arrow). (B) STIR axial MRI image showing associated skin thickening with the lesion (arrow), and epsilateral axillary lymphadenopathy (notched arrow) together with diffusion restriction seen in ADC image (arrow in C) raise the malignant suspicion. (D) MIP for post-contrast axial T1w fat sat image shows grade I vascularity as relatively engorged local vascularity with slightly increased caliber. (E) DCE curve shows continuously raising curve consistent with benign nature of the swelling, chronic inflammatory process (granulomatous mastitis) was the pathological result of US guided biopsy.

Figure 5 (A and B) Right breast ML mammogram and US breast show central nodule which was suspected to be malignant due to dense acoustic shadow in US. (C) MRI T1 postcontrast T1 fat sat showed avid post-contrast enhancement of the lesion. MIP images of early (D) and delayed (E) phases of dynamic postcontrast sequence show low vascular score (1) with no new angiogenesis. This together with benign continuously raising DCE curve (F) had supported the benign radiological diagnosis. (Pathologically proven traumatic fat necrosis with fibrous scar).

Figure 6 (A) CC mammogram showing left breast pathologically proven malignant nodule. (B) DCE curve of the lesion seen in A showing early peak with rapid washout (malignant curve). (C) Plateau curve for another malignant lesion of the right breast (D). (E) CC mammogram showing right breast pathologically proven benign nodule, (F) DCE curve demonstrating continuos rising enhancement (benign enhancement curve), (G) table of sequential ROI measures in DCE study showing continuos raising figures.

Table 2 Results of the MRI examinations.

Score range	No. of patients	B.P. benign	B.P. malignant	Sensitivity %	Specificity %	PPV %	NPV %
<5	36	30	6	93.1	90.9	96.4	83.3
⩾5	84	3	81
Total	120	33	87

No.: number, B.P: biopsy proven.

Sensitivity: true positive/true positive + false negative.

Specificity = true negative/true negative + false positive.

PPV (positive predictive value) = true positive/true positive + false positive.

NPV (negative predictive value) = true negative/true negative + false positive.