Abbreviated and Ultrafast Dynamic Contrast-enhanced (DCE) MR Imaging

Abstract

The early detection and treatment of breast cancer is extremely important for extending patients’ outcomes. Breast MRI has high sensitivity for the detection of breast cancer and plays an important role in breast cancer diagnosis and treatment, but conventional dynamic contrast-enhanced (DCE) MRI may be too time-consuming for breast cancer screening purposes. Abbreviated MRI is a technique that can be applied within a short time, as usually only the pre-contrast and first post-contrast images from the dynamic study or additional T2-weighted imaging are used. Abbreviated MRI may thus be suitable for breast cancer screening. In addition, its diagnostic performance for differentiating benign and malignant breast lesions is superior to that of breast tomosynthesis and comparable to that of conventional DCE MRI. The usefulness of abbreviated MRI for patients with a history of breast cancer and in clinical settings has been described, but the specificity of abbreviated DCE MRI is slightly lower than that of conventional DCE MRI. Ultrafast DCE MRI is a technique that obtains kinetic information by capturing multiple time phases in a short time scan in the very early phase after the injection of contrast material. Various parameters, including the maximum slope and time to enhancement can be used to evaluate kinetic information. Based on this kinetic information, ultrafast DCE MRI can differentiate between benign and malignant breast lesions. Since background parenchymal enhancement (BPE) is weak in the very early phase after a contrast media injection, ultrafast DCE MRI is also useful for identifying lesions in patients with marked BPE on conventional DCE MRI. In addition, ultrafast DCE MRI is useful for predicting the prognostic marker status of breast cancer, assessing the effectiveness of neoadjuvant therapy, examining MRI-detected lesions before surgery, and morphological assessments.

Introduction

Breast cancer is the most common cancer affecting women and the fourth leading cause of death among women,¹ and the early detection and treatment of breast cancer are thus extremely important. Breast MRI has high sensitivity for the detection of breast cancer and plays an important role in breast cancer diagnosis and treatment, including staging,^2–5 evaluating chemotherapy responses,^6,7 screening among woman at high risk,^8,9 evaluating patients with axillary lymph node metastasis with an unknown primary tumor,¹⁰ and problem-solving along with mammography and ultrasound examinations.^5,11 Although breast MRI has high sensitivity for breast cancer diagnoses, its specificity is not as high as its sensitivity.¹² Therefore, in breast MRI, breast cancer is diagnosed both by evaluating morphology using high-spatial-resolution images after contrast-enhancement and by evaluating kinetic information based on time-intensity curves obtained from a dynamic contrast-enhanced (DCE) study.

Early breast MRI studies revealed that breast cancer exhibits peak enhancement early after the injection of contrast material¹³ and a washout pattern late after the contrast material’s injection.¹⁴ A kinetic evaluation based on a dynamic study thus plays an important role in diagnosing breast cancer. According to the Breast Imaging-Reporting and Data System (BI-RADS) published by the American College of Radiology (ACR),¹⁵ which is widely used worldwide and lists findings and terminology related to breast imaging, the ‘initial phase’ is defined as the first 2 mins after the injection of contrast material or until peak enhancement is reached. The ‘delayed phase’ is usually set up to 5–10 mins after the contrast material injection. The fast and washout pattern is a typical malignancy pattern and it takes 4 to 5 mins or more after injection to get this typical malignant washout pattern. However, this length of time may be too long for screening purposes. In addition, if the number of screening MRI increases, long scanning time may limit the number of MR scans per day. In these circumstances, abbreviated MRI was devised as an efficient method for screening MRI. In addition, ultrafast DCE MRI has been developed, which can obtain kinetic information in a short scan time. This review will introduce the basics and recent developments of both abbreviated MRI and ultrafast DCE MRI.

Abbreviated MRI

Abbreviated MRI was designed to shorten and improve the efficiency of imaging and interpretation times in screening MRI. Initially, the purpose of abbreviated MRI was specifically the detection of breast cancer. In the first report by Kuhl et al., the subjects included women with a family history of breast cancer as well as those with dense breasts and those with a personal history of breast cancer.¹⁶ As shown in Fig. 1, usually only pre-contrast and first post-contrast images from the dynamic study are used in abbreviated MRI. For interpretation, these images, subtracted images, and the maximum intensity projection (MIP) of subtracted images (Fig. 1) or MIP images alone¹⁶ have been used. The acquisition times have been 17 mins for the full diagnostic protocol (FDP) and 3 mins for the abbreviated protocol (AP). With the use of complete AP images, the interpretation time was 28 sec; with the use of only MIP images, the interpretation time was only 3 sec.¹⁶

Fig. 1.

The conventional DCE MRI, abbreviated MRI, and ultrafast DCE MRI protocols. DCE, dynamic contrast-enhanced; DWI, diffusion weighted image; MIP, maximum intensity projection; T1WI, T1-weighted image; T2WI, T2-weighted image.

The use of FDP has provided the following values: sensitivity, 100%, specificity, 94.3%; positive predictive value (PPV), 24.4%; and negative predictive value (NPV), 100%. The corresponding values achieved with AP are: sensitivity, 100%; specificity, 93.9%; PPV, 23.4%, and NPV, 100%. The use of MIP images alone for interpretation has shown 90.0% sensitivity and an NPV of 99.8%.¹⁶ Abbreviated MRI is thus considered suitable for screening because it has demonstrated almost the same breast cancer detection ability as that of full-protocol MRI and can shorten the acquisition time.

Diagnostic performance and scan protocol

As summarized in Table 1, investigations of the diagnostic performance of the AP and the FDP have described 81.8%–100% sensitivity and 45%–97.2% specificity for the AP, and 81.8%–100% sensitivity and 52%–97.4% specificity for the FDP.^16–28 These diagnostic performance values are comparable, but in many studies the specificity of the AP was found to be slightly lower than that of the FDP. Some of these studies were conducted on screening cohorts,^{16,19,20,25–30} and others were performed with enriched cohorts.^{17,18,21–24} The scan time of the FDP ranged from 15 to 40 mins, and that of the AP was 3–13 mins.^{16–21,24–28,30} In most of the studies, the AP was <10 mins. The reading times for the AP and the FDP were 0.5–2.98 mins and 1.0–6.6 mins, respectively.^{16–21,24–28,30}

Table 1.

Comparison of the diagnostic performance of the AP and the FDP

Study	1.5T or 3T	Study design	Population	Added T2WI?	AP		FDP
					Sen (%)	Spe (%)	Sen (%)	Spe (%)
Kuhl16	1.5	Prospective	Screening	No	100	93.9	100	94.3
Grimm17	1.5	Retrospective	Enriched	Yes	86 (AP1)	52 (AP1)	95	52
					89 (AP2)	45 (AP2)
Romeo18	1.5	Retrospective	Enriched	No	99	93	97	95
Chen19	3	Retrospective	Screening	No	93.8	88.3	100	94.9
Chen20	3	Retrospective	Screening	No	92.9 (AP1)	86.5 (AP1)	100	96.8
					100 (AP2)	95 (AP2)
Moschetta21	1.5	Retrospective	Enriched	Yes	89	91	89	92
Machida22	3	Retrospective	Enriched	No	87.1	91.7	87.1	90.3
					93.5	83.4	96.8	89.7
Petrillo23	1.5	Retrospective	Enriched	No	99.5	75.4	99.5	77.1
Oldrini24	1.5	Retrospective	Enriched	Yes	93.1	60.4	93.1	60.4
					93.1	60.4	93.1	58.3
Panigrahi25	1.5	Prospective	Screening	No	81.8	97.2	81.8	97.4
Seppala26	1.5	Retrospective	Screening	No	69.6	77.9	83	83
Dialani27	1.5	Retrospective	Screening	No	93	93	100	86
Naranjo28	3	Retrospective	Screening	Yes	92 (AP1)	72 (AP2)	94	80
					90 (AP1)	83 (AP2)

AP1: first post-contrast subtracted (FAST) + T2WI, AP2: FAST+ 2nd phase of post contrast subtracted image + T2WI in the study by Grimm et al. AP1: FAST, AP2: FAST + DWI in the Chen et al. study. AP1: FAST, AP2: FAST + T2WI in the Naranjo et al. study. The studies by Machida et al. and Oldrini et al. provided the results of 2 readers. AP, abbreviated protocol; FDP, full diagnostic protocol; Sen, sensitivity; Spe, specificity; T2WI, T2-weighted image.

Abbreviated MRI can be easily performed with a 1.5T scanner or a 3T scanner. However, AP was not strictly defined. Some studies of APs used only subtraction images of pre-contrast and a single post-contrast (first post-contrast subtracted [FAST]) image plus the MIP images (Fig. 1),^{16,19,20,22,23,25,26,30} while other AP studies added T2-weighted images (T2WI) to these (Fig. 1).^{17,21,24,27,28} In 1 investigation, subtraction images of pre- and 2 post-contrast phases (called a ‘simplified breast protocol’) were used.¹⁸ Some research groups have compared the diagnostic performance of several APs. For example, Grimm et al. compared the diagnostic performance of AP1 (using T2WI, 1 pre- and 1 post-contrast phase) and AP2 (using T2WI, 1 pre- and 2 post-contrast phases), and they observed no significant differences in sensitivity or specificity between AP1 and AP2.¹⁷ In abbreviated MRI, the use of only 1 phase of postcontrast imaging may be sufficient, apart from the addition of T2WI.

Naranjo et al. compared the diagnostic performances of APs with and without T2WI, and they reported that (i) the performance of the AP without T2WI was suboptimal, and (ii) the performance provided by the AP with T2WI was comparable to that of FDP.²⁸ In abbreviated MRI, the addition of T2WI may thus help improve the diagnostic performance. On the other hand, Chen et al. compared the diagnostic performance of APs with and without the use of diffusion weighted imaging (DWI).²⁰ They reported that the sensitivity values of the APs with and without DWI were 100% and 92.9%, respectively. The specificity values of the APs with and without DWI were 95.0% and 86.5%, respectively. Compared to the AP without DWI, the AP with DWI had superior sensitivity and specificity.²⁰ DWI may thus help improve the diagnostic performance of abbreviated MRI.

Two meta-analyses summarize many of these studies;^31,32 Baxter et al. reported the pooled diagnostic performance of an AP and the FDP in each study group conducted with a screening cohort and an enriched cohort.³² The following performance values were identified: for the screening cohort, the pooled sensitivity was 90%, the specificity was 92%, and the area under the curve (AUC) was 0.94 with the AP, and the corresponding values for the FDP were 92%, 95%, and 0.97. For the enriched cohort, the pooled sensitivity was 93%, the specificity was 83%, and the AUC was 0.94 with the AP, and the corresponding values for the FDP were 93%, 84%, and 0.95.³² There were no significant differences in sensitivity or specificity between the uses of an AP or the FDP in each subgroup. In the other meta-analysis, Geach et al. reported that the overall sensitivity and specificity values for APs with the FDP as reference standard were 94.8% and 94.6%, respectively.³¹ There were also no significant differences in sensitivity or specificity between the uses of an AP or the FDP.

Kwon et al. identified no significant difference in diagnostic performance after 3 consecutive years of screening using abbreviated MRI, and they noted that the diagnostic performance was maintained.³³ They also reported that missed cancer at abbreviated MRI tended to be smaller than detected cancer.³³ Liu et al. observed that after training, the diagnostic performance and the inter-reader agreement of abbreviated MRI for small lesions were improved, although they were still inferior to those of the FDP.³⁴

Comparison with breast tomosynthesis

Mammography is the only breast cancer screening method that has been confirmed to reduce breast cancer’s mortality rate. However, the sensitivity and specificity of mammography are decreased in dense breasts. Breast tomosynthesis is expected to improve the sensitivity and specificity in screening of dense breasts. The diagnostic performance of tomosynthesis and that of abbreviated MRI in dense breasts has been compared.^35,36 Comstock et al. evaluated these 2 modalities’ screening performance in dense breasts, and they reported that tomosynthesis had 39.1% sensitivity and 97.4% specificity, whereas abbreviated MRI had 95.7% sensitivity and 86.7% specificity.³⁵ In addition, abbreviated MRI showed a higher cancer detection rate than tomosynthesis and was useful in dense breasts in a screening cohort.

Ramli Hamid et al. conducted a similar study but in an enriched cohort (screening and diagnostic populations). Tomosynthesis provided 98.5% sensitivity and 34.6% specificity; abbreviated MRI showed 98.5% sensitivity and 43.6% specificity.³⁶ The diagnostic performance of abbreviated MRI may thus be comparable to that of tomosynthesis in diagnostic populations. Kim et al. compared the diagnostic performance of tomosynthesis and abbreviated MRI in a series of women with a personal history of breast cancer (PHBC),³⁷ and their analyses revealed 54.6% sensitivity and 97.6% specificity for tomosynthesis, and 100% sensitivity and 96.5% specificity for abbreviated MRI.³⁷ Abbreviated MRI thus showed higher sensitivity without a significant decrease in specificity in women with a PHBC. These findings demonstrated that in some situations, abbreviated MRI provides superior diagnostic performance compared to tomosynthesis.

Diagnostic performance in screening for women with a PHBC

Many early studies of abbreviated MRI included a variety of patient populations or were limited to women at high risk for breast cancer. In contrast, women with a PHBC are at risk of ipsilateral recurrence and contralateral breast cancer in which the lifetime risk of breast cancer is >15%–20%.³⁸ As described in Table 2, several recent studies indicate the usefulness of abbreviated MRI for women with a PHBC.

Table 2.

Diagnostic performance of abbreviated MRI in patients with a PHBC

Study	AP		Other methods (for comparison)
	Sen (%)	Spe (%)	Sen (%)	Spe (%)
Choi29	100	89	None	None
Ha39	92.2–94.9	97.6–98.6	97.9–98.3	97.9–98.3
			(FDP)	(FDP)
Park40	70	98	10 (FDP)	96.9 (FDP)
Baek42	68.8	97.0	18.8 (MMG)	99.5 (MMG)
			25.0 (US)	98.2 (US)
An41	100	96	None	None
Kim44	100	93	69 (FDP)	86 (FDP)
Kim37	100	96.5	54.6 (Tomo)	97.6 (Tomo)
Kim43	90	88.6	90 (AP + Ultrafast)	95.3 (AP + Ultrafast)

AP, abbreviated protocol; FDP, full diagnostic protocol; MMG, mammography; PHBC, personal history of breast cancer; Sen, sensitivity; Spe, specificity; Tomo, tomosynthesis; US, ultrasound.

In several investigations of women with a PHBC, the diagnostic performance of abbreviated MRI was represented by 68.6%–100% sensitivity and 88.6%–98.6% specificity.^{29,37,39–44} Studies comparing abbreviated MRI with the FDP have shown that its diagnostic performance is comparable,^39,40,44 and comparisons of abbreviated MRI with mammography, tomosynthesis, and ultrasound demonstrated that the diagnostic performance of AP is superior.^37,42 Kim et al. observed that specificity was improved by the addition of an ultrafast protocol to an AP.⁴³

Abbreviated MRI in diagnostic settings

The usefulness of abbreviated MRI in diagnostic settings has been evaluated. Gweon et al. reported that the application of abbreviated MRI increased the PPV for suspicious microcalcification on screening mammography.⁴⁵ Compared with the FDP, the diagnostic performance of abbreviated MRI was comparable in evaluations of the tumor extent, nipple areolar complex invasion, and regional lymph node metastasis in preoperative staging.^46–51 Compared with the use of MIP images alone, the addition of FAST images appears to have a higher diagnostic ability, and the addition of DWI appears to further improve diagnostic ability.^47,51 On the other hand, Shiraishi et al. reported that the tumor extent evaluation of pure ductal carcinoma in situ (DCIS) by abbreviated MRI was inferior than that of the FDP.⁵² Several research groups have assessed the effectiveness of neoadjuvant chemotherapy (NAC) for breast cancer:^53–55 Yirgin et al. reported that the evaluation of the local extent of the tumor by abbreviated MRI was slightly inferior compared to that provided by the FDP,⁵³ whereas Tang et al. and Dornelas et al. reported no significant differences in the evaluation of the response to NAC compared to FDP.^54,55 Further research regarding the application of abbreviated MRI to diagnostic settings is therefore merited.

Ultrafast DCE MRI

The usefulness of abbreviated MRI in screening has been recognized, but it has been pointed out that its specificity is slightly lower than that of the FDP, although significant differences have not been identified. In the first abbreviated MRI report, approximately one-third of the cases classified as BI-RADS category 2 by the FDP were classified as category 3 by abbreviated MRI.¹⁶ Because follow-up after 6 months is recommended for category 3 cases, there is a risk that the number of unnecessary follow-up cases will increase with abbreviated MRI. As noted earlier, it was also reported that the number of cases in which cancer was missed by abbreviated MRI tended to be smaller than that of detected cancer,³³ and even with training, the ability for diagnosing small lesions was lower than that of the FDP.³⁴ Just as a kinetic evaluation is important in addition to the morphological evaluation in conventional DCE (i.e., the FDP), the addition of a kinetic evaluation is thought to be useful, even in short-time scan imaging. Ultrafast DCE MRI was developed to address these circumstances.⁵⁶ It is a technique that obtains kinetic information by capturing multiple time phases in a short time scan during the very early phase after contrast administration (Fig. 1) and it covers the whole breast.^57,58

Techniques for obtaining ultrafast images

To obtain ultrafast images, a technique that has high temporal resolution while maintaining a certain degree of spatial resolution is required. Mann et al. used a view-sharing technique to acquire ultrafast images in the first report concerning ultrafast dynamic imaging.⁵⁶ In the view-sharing technique, outer k-space sampling data are shared with data originating from other timepoint phases.⁵⁶ This technique has been adopted in many studies.^{24,43,56,59–85} Several specific sequences can be used with the view-sharing technique (TWIST, KWIC, DISCO, TRICS, and 4D-TRAK), which vary depending on the vendor.^57,58 However, data contamination may cause artifacts throughout all data-sharing phases in the view-sharing technique.

Compressed sensing (CS), another technique used for ultrafast DCE MRI,^86–100 uses the sparsity of a signal to recover the original signal from a signal that has been decimated, by applying a specific sampling technique; the use of CS enables higher spatio-temporal resolutions and better motion robustness.¹⁰¹ Iterative reconstruction is needed for CS. Sagawa et al. reported that at least 15 iterations were needed for semi-quantitative parameters of CS-derived ultrafast DCE MRI.¹⁰² The CS technique was not widely used in the past, but its application has become more widespread in recent years. There are also recent reports concerning the use of the golden-angle radial sparse parallel (GRASP) technique, which is a type of CS technology.^103,104

Parallel imaging is also used.^105–114 However, it is difficult to set the parameters for this method, and if the parameters are not set properly, the time resolution and slice thickness may be somewhat inferior to those of other methods. To improve the time resolution, high-performance equipment is required, and many studies have used a 3T MRI system. However, some studies used 1.5T MRI or both 1.5T and 3T MRI.^{24,74,76,77,79,81,115}

Acquisition parameters are varied and not standardized. As indicated in Table 3, the temporal resolution has varied from 1.7 to 12 sec, the slice thickness has varied from 0.8 to 4 mm, and the in-plane spatial resolution has varied from 0.68 × 0.68 to 1.2 × 2.7 mm. The standardization of acquisition parameters is necessary in order for ultrafast MRI to become widespread. The magnetic field, acquisition technique, temporal resolution, in-plane spatial resolution, and slice thickness of ultrafast DCE studies are listed in Table 3.

Table 3.

The magnetic field, acquisition technique, temporal resolution, in-plane spatial resolution, and slice thickness of the ultrafast DCE studies

References	Magnetic field	Acquisition technique	Temporal resolution (sec/phase)	In-plane spatial resolution (mm)	Slice thickness (mm)
Mann56	3T	VS	4.32	1.0 × 0.9	2.5
Pineda106	3T	PI	6.9–9.9	1.5 × 1.5	3
Abe105	3T	PI	7	1.5 × 1.5	3
Onishi86	3T	CS	3.65	0.9 × 0.9	2.5
Mus59	3T	VS	4.32	1.0 × 0.9	2.5
Milenkovic´60	3T	VS	4.3	0.94 × 0.94	2.5
Mori107,112	3T	PI	3	1.09 × 1.66	4
van Zelst61	3T	VS	4.3	1.9 × 0.9	2.5
Cover62	3T	VS	4.26	N/A	2.2
Wu108	3T	PI	3.4–4.1	1.5 × 1.5	4
			1.7–3.5
Goto63	3T	VS	5.3	0.9 × 0.9	2.5
Honda87,89,92	3T	CS	3.7	0.9 × 0.9	2.5
Onishi64,66	3T	VS	2.7–7.1	1.6 × 1.6	1.6
Ohashi65	3T	VS	2.5	0.94 × 0.94	2.5
Shin67	3T	VS	3.8	1.11 × 1.24	1.0
			4.5	0.94 × 0.94	1.0
Kim68	3T	VS	3.8	1.1 × 1.1	1.0
			4.5	1.0 × 1.0	1.0
Yamaguchi109	3T	PI	8.3	0.9 × 0.9	2.5
Mori110	3T	PI	7	1.5 × 1.5	3
Peter69	3T	VS	4.9	0.9 × 0.9	2.5
Ohashi88,95,96	3T	CS	3.7	0.94 × 0.94	2.5
Jeong70	3T	VS	8.1	1.0 × 1.0	1.0
Lee71	3T	VS	6.5	1.6 × 1.6	1.6
Ayatollahi73	3T	VS	4.3	1.0 × 0.9	2.5
Pelissier74	1.5T	VS	7.1	1.38 × 1.17	2
Kim90	3T	CS	4.7	0.8 × 1.1	0.9
Kato111	3T	PI	3	1.09 × 1.66	4
Heo75	3T	VS	6.7	1.6 × 1.6	1.6
Jing76,79,81	3T	VS	4.3	0.91 × 0.91	3
	1.5T		5.2	0.68 × 0.68	3
Ramtohul77	1.5T	VS	5.9–6.0	1.1 × 1.6	4
Yamaguchi91,99	3T	CS	2.9	0.9 × 0.9	2.5
Kim78	3T	VS	6.5	0.78 × 0.78	3
Kim43	3T	VS	4	0.9 × 0.9	1
Ren113,114	3T	PI	3–9	1.5 × 1.5	3–4
Nissan104	3T	CS	6.1	1.3 × 1.3	1.5
Ramli Hamid80	3T	VS	3.7	1.0 × 0.9	2.5
Cao93,94,98	3T	CS	4.5	0.9 × 0.9	2.5
Milon115	3T	N/A	7.28	1.2 × 2.6	2.8
	1.5T			1.2 × 2.7	2
Miceli82	3T	VS	5	1.2 × 0.94	2.5
Lyu84	3T	VS	7	N/A	1.2
Kim97	3T	CS	4.5	0.7 × 0.7	0.8
Luo83	3T	VS	12	N/A	2.5
Xie85	3T	VS	4.7	N/A	2.1
Huang100	3T	CS	4.5	0.9 × 0.9	2.5

CS, compressed sensing; PI, parallel imaging; VS, view-sharing.

Parameters of ultrafast DCE MRI

Breast cancer is generally a hypervascular and hyperperfused tumor, and it is strongly enhanced in the early phase after the injection of a contrast agent. Ultrafast MRI captures multiphase images from the start of contrast enhancement to this early phase, and various semi-quantitative parameters have been devised to evaluate these kinetics (Fig. 2). The most commonly used are the maximum slope (MS) and the time to enhancement (TTE). The MS (%/sec), defined as the slope of the tangent at the steepest part of the time-intensity curve,⁵⁶ reflects the first pass of the contrast medium through the lesion and perfusion. In general, MS values are higher in breast cancer than in benign lesions.

Fig. 2.

Schema of each parameter of ultrafast DCE MRI. AUC, area under the curve; BAT, bolus arrival time; DCE, dynamic contrast-enhanced; MS, maximum slope; TTE, time to enhancement; TTP, time to peak; WIS, wash in slope.

The TTE (sec) is defined as the time point at which the lesion starts to enhance minus the time point at which the aorta starts to enhance.⁵⁹ Breast cancer with abundant blood flow becomes clearly visible early after a contrast material injection, and the TTE is shorter in the tumor than in benign lesions. Parameters that are similar to TTE include the time of arrival (TOA) and the bolus arrival time (BAT). The TOA is defined as the time of the arrival of the contrast medium from the mammary arteries to the lesion.^62,106 It has been reported that the TOA values of malignant lesions were significant shorter than that of benign lesions.¹⁰⁶ The BAT is defined as the time from the start of the contrast material injection to the arrival of the tracer bolus at a lesion.^64,66 In an investigation by Onishi et al., sub-centimeter-sized breast cancer showed significant shorter BAT values compared to benign lesions.⁶⁴ Aggressive breast cancer also showed significant shorter BAT compared to less-aggressive cancer.⁶⁶

The kinetic AUC, which was developed in the early days of ultrafast MRI, is the area under the curve of the time-intensity curve of ultrafast DCE MRI. A kinetic AUC is calculated from zero to each timepoint.¹⁰⁵ Abe et al. observed that the kinetic AUC of malignant lesions was significantly larger than that of benign lesions.¹⁰⁵ The time to peak (TTP) and the wash in slope (WIS) are parameter that are relatively new. The TTP (sec) is defined as the time required for the contrast medium to arrive at the lesion and reach peak intensity.⁷⁷ The WIS (%/sec) is defined as the peak signal intensity (SImax) divided by the TTP.⁷⁷ Ramtohul et al. proposed that the WIS is useful for predicting a pathologic complete response (pCR) of breast cancer after neoadjuvant chemotherapy.⁷⁷

A unique parameter of ultrafast DCE MRI is the arterial and venous interval (AVI), which is defined as the time interval between the start of arterial enhancement and the beginning of venous enhancement.⁸⁶ The AVI of breast cancers is shorter than that of benign lesions, reflecting the impaired tumor vasculature and shunt formation of breast cancer.^86,87

The quantitative pharmacokinetics parameters (Ktrans, Kep, and Ve) are also used for the evaluation of ultrafast DCE MRI.^{85,94,108,114} The empirical mathematical model (EMM) is used to fit the uptake and washout of contrast media on DCE MRI.¹¹⁶ Pineda et al. used a truncated EMM to fit ultrafast DCE MRI.¹⁰⁶ Five parameters can be obtained from a truncated EMM: the upper limit of the signal intensity (A), the rate of signal increase (α), the initial slope of the kinetic curve (Aα), the initial AUC, and the time of initial enhancement.¹⁰⁷ These parameters significantly correlated with the microvessel density (MVD) of breast cancer in a study by Mori et al.¹⁰⁷

Attempts to separate and visualize tumor-related vessels have been described.^58,108 There are also several reports of the incorporation of artificial intelligence (AI), e.g., deep-learning and radiomics techniques, into analyses of lesion detection, the differentiation of benign and malignant masses, the optimization of image acquisition, shortened reading times, the diagnosis of extensive intraductal components, and the prediction of intrinsic subtypes.^{60,73,76,79,81,84,100}

Diagnostic performance for differentiating benign and malignant breast lesions

As demonstrated in Table 4, many research groups have reported that the diagnostic performance of ultrafast DCE MRI for differentiating benign and malignant breast lesions is comparable to or superior to that of conventional DCE MRI. A systematic review and meta-analysis of the diagnostic performance for differentiating benign and malignant breast lesions was published in 2024¹¹⁷ and revealed the following: with the use of the best ultrafast DCE MRI parameter of each study, the pooled sensitivity was 83%, the specificity was 77%, and the AUC of the receiver operating characteristic (ROC) curve was 0.876.¹¹⁷ With the use of the MS individually, the pooled sensitivity was 80%, the specificity was 77%, and the AUC of the ROC curve was 0.865, whereas using the TTE individually, the pooled sensitivity was 71%, the specificity was 80%, and the AUC of the ROC curve was 0.57.¹¹⁷ The authors of that meta-analysis concluded that ultrafast DCE MRI as a stand‑alone technique has high accuracy in discriminating benign from malignant breast lesions.¹¹⁷ Representative cases of breast cancer and benign lesion on ultrafast DCE MRI are provided in Fig. 3.

Table 4.

Studies directly comparing the diagnostic performance of ultrafast DCE MRI and conventional DCE MRI for differentiating benign and malignant breast lesions

Reference	Ultrafast DCE MRI		Conventional DCE MRI
	Parameter	Results	Parameter	Results
Mann56	MS	AUC: 0.829	BI-RADS	AUC: 0.692
Abe105	Kinetic AUC	Sen: 85%	SER	AUC: 0.88
		Spe: 79%
		AUC: 0.87
Mus59	TTE	Sen: 90.9%–93.9%	BI-RADS	Sen: 90.9%–95%
		Spe: 77.1%–79.1%		Spe: 51.0%–53.1%
		AUC: 0.8–0.86		AUC: 0.7%–0.71
Oldrini24	Afferent vessel + abbreviated MRI	Sen: 93.1%	BI-RADS	Sen: 93.1%
		Spe: 70.8%–83.3%		Spe: 58.3%–60.4%
van Zelst61	MS, TTE	Sen: 84%	BI-RADS	Sen: 86%
		Spe: 82%		Spe: 76%
		AUC: 0.89		AUC: 0.89
Goto63	MS + TTE + BI-RADS	AUC: 0.864 (mass)	BI-RADS	AUC: 0.823 (mass)
		AUC: 0.923 (NME)		AUC: 0.865 (NME)
Ohashi65	MS	AUC: 0.81	BI-RADS	AUC: 0.96
Honda87	MS, TTE, AVI	AUC: 0.76 (MS, AVI)	BI-RADS	AUC: 0.69
		AUC: 0.78 (TTE)
Mori110	EMM	AUC: 0.71–0.81	Morphologic score	AUC: 0.81–0.89
			SER

AUC, area under the curve; AVI, arterial and venous interval; BI-RADS, Breast Imaging-Reporting and Data System; EMM, empirical mathematical model; MS, maximum slope; NME, non-mass enhancement; Sen, sensitivity; SER, signal enhancement ratio; Spe, specificity; TTE, time to enhancement.

Fig. 3.

A 67 year-old woman with invasive carcinoma of no special type at the left breast (a, b) and a 55 year-old woman with a benign lesion (mastopathy) at the left breast (c, d). (a, c) MIP images obtained with ultrafast DCE MRI. (b, d) Time-intensity curves of ultrafast DCE MRI. The breast cancer shows a higher MS (30.56) and shorter TTE (2.9 sec) compared to the benign lesion (MS: 8.74, TTE: 11.6 sec). DCE, dynamic contrast-enhanced; MIP, maximum intensity projection; MS, maximum slope; TTE, time to enhancement.

The MS and TTE values in benign and malignant lesions have varied among studies (Table 5). This variation may be due to the study design and/or the type of cohort examined, and differences in acquisition methods may also be a factor. It seems that with the use of higher time resolution or the CS technique, higher MS values and shorted TTE values are obtained.

Table 5.

The MS and TTE values in benign and malignant lesions

MS (%/sec)
Reference	Benign	Malignant	Time resolution (sec)	Technique
Goto63	5.9	9.8	5.3	VS
Ohashi65	12.4	25.1	3.75	VS
Lee71	6.8	11.8	6.5	VS
Honda87	18.4	29.3	3.7	CS
Pelissier74	3.6	11.6	7.1	VS
Ramli Hamid80	5.45	13.27	3.7	VS
TTE (sec)
Reference	Benign	Malignant	Time resolution (sec)	Technique
Goto63	14	9.9	5.3	VS
Honda87	12	7	3.7	CS
Ramli Hamid80	10.11	7.67	3.7	VS
Yamaguchi91	11.07	6.43	2.9	CS

CS, compressed sensing; MS, maximum slope; TTE, time to enhancement; VS, view-sharing.

Several research groups have noted that the diagnostic performance for differentiating benign and malignant breast lesions was improved by combining the parameters of ultrafast DCE MRI with other parameters. For example, Goto et al. reported that the AUCs of TTE + MS combined with BI-RADS (0.864 in mass and 0.923 in non-mass enhancement [NME]) were better than those obtained with BI-RADS alone (0.823 in mass and 0.865 in NME).⁶³ Onishi et al. reported that an MS + BAT + age classification model resulted in a higher AUC compared to those obtained with an MS + BAT model or an age model (0.846, 0.704, and 0.738, respectively) in sub-centimeter breast lesions.⁶⁴ In an investigation by Peter et al., a classification model using the peak enhancement of ultrafast DCE MRI and the apparent diffusion coefficient (ADC) value of DWI provided performance that was equal to that of a model using conventional DCE MRI and the ADC (93.3% vs. 93.3% sensitivity, 91.3% vs. 87.0% specificity).⁶⁹ Ohashi et al. observed that an MS + ADC + age classification model provided a higher AUC than MS, ADC and age models (0.9, 0.73, 0.74, and 0.87, respectively).⁸⁸ A combination of ultrafast DCE-MRI semi-quantitative multiparameters (the MS, TTP, TTE, and kinetic AUC) and the ADC showed better classification performance than the ultrafast DCE-MRI semi-quantitative multiparameters and quantitative multiparameters (Ktrans, Kep, and Ve) in a study by Cao et al.⁹⁴ Milon et al. reported that an abbreviated protocol with ultrafast DCE-MRI + ADC provided superior performance compared to a conventional protocol, Conventional + ADC and abbreviated protocol with ultrafast.¹¹⁵

The relationship between ultrafast DCE MRI parameters and prognostic markers of breast cancer

Many studies have investigated the relationship between ultrafast DCE MRI parameters and prognostic markers of breast cancer (Table 6). The kinetic ultrafast DCE MRI parameters are associated with both the presence of an invasive component and prognostic markers of breast cancer because tumor vascular formation is closely linked to the tumor growth. Invasive breast cancer has higher MS and shorter TTE or BAT values.^{66,67,87,109,118} Invasive breast cancer with higher malignant potentials, i.e., invasive breast cancer with a larger size, higher Ki-67 index value, higher histological or nuclear grade, negative hormone receptor status, positive human epidermal growth factor 2 (HER2) status, triple negative (TN) status, axillary lymph node metastasis, etc., tends to have higher MS and shorter TTE and BAT values compared to breast cancer with lower malignant potentials.^{63,66,67,74,87,90,95,109,118} A recent examination of the relationship between tumor-infiltrating lymphocytes (TILs) and the MS or TTE is available.⁹⁹

Table 6.

The relationship between major parameters of ultrafast DCE MRI and prognostic markers of breast cancer

Marker	MS	TTE	BAT	References
IC or DCIS	IC > DCIS	IC < DCIS	IC < DCIS	66 , 67 , 87 , 109 , 118
Size	Large > Small	Large < Small		67 , 74 , 90
Ki-67	High > Low	High < Low		63 , 67 , 74 , 90 , 109
Histological or nuclear grade	High > Low	High < Low	High < Low	63 , 66 , 67 , 74 , 90 , 109
HR	Negative > Positive	Negative < Positive	Negative < Positive	63 , 74 , 90
HER2		Positive < Negative		67 , 90
Intrinsic subtype	TN + HER2 + Luminal B > Luminal A + ILC	TN < non-TN	TN or HER2 < Luminal	66 , 74 , 90 , 95
Histological subtype	IDC > ILC or Mucinous ca.	IDC < ILC or Mucinous ca.	IDC < ILC	66 , 87
TILs	High > Low	High < Low		99
Axillary lymph node metastasis	Positive > Negative			74 , 109

BAT; bolus arrival time; ca., carcinoma; DCIS, ductal carcinoma in situ; HER2, human epidermal growth factor receptor 2; HR, hormone receptor; IC, invasive carcinoma; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; MS, maximum slope; TILs, tumor-infiltrating lymphocytes; TN, triple-negative; TTE, time to enhancement.

Ohashi et al. applied a model combining ultrafast DCE MRI and early-phase DCE-MRI parameters, and they reported that this model was useful for diagnosing TN breast carcinoma.⁹⁵ A multiparametric MRI model that included ultrafast DCE MRI, conventional DCE MRI, magnetic resonance spectroscopy, diffusion kurtosis imaging, and intravoxel incoherent motion was useful for predicting intrinsic subtypes in an investigation by Huang et al.¹⁰⁰ Several groups have also reported that ultrafast MRI is useful for predicting postoperative upgrade in DCIS cases diagnosed by biopsy.^75,82,97,112

The relationship between background parenchymal enhancement and lesion visualization on ultrafast DCE MRI

Background parenchymal enhancement (BPE) sometimes affects the interpretation of breast MRI findings. In cases of severe BPE, it can sometimes be difficult to identify the breast lesions on conventional DCE MRI. Honda et al. observed that the BPE was lower on ultrafast DCE MRI compared to conventional DCE MRI,⁸⁹ and other studies revealed that the lesion detectability or conspicuity on ultrafast DCE MRI was superior than on conventional DCE MRI in patients with higher BPE on conventional DCE MRI and those with a premenopausal status, NME, or a shorter TTE.^68,89 Nissan et al. reported that the lesion conspicuity of pregnancy-associated breast cancer was improved on ultrafast DCE MRI due to reduced lactation BPE.¹⁰⁴ Thus, ultrafast DCE MRI is useful for identifying and evaluating breast lesions in cases of severe BPE on conventional DCE MRI.

Yamaguchi et al. noted that the although the TTE interval between cancer and parenchymal foci in patients with a premenopausal status was shorter than that in patients with a postmenopausal status, the TTE was useful to differentiate breast cancer from benign lesions and normal breast parenchymal foci in both pre- and postmenopausal status with the use of different cutoff values.⁹¹ Figure 4 depicts a representative case in which BPE is reduced on ultrafast DCE.

Fig. 4.

A 50 year-old woman with invasive carcinoma of no special type at the left breast. (a) Early phase of conventional DCE MRI, (b) ultrafast DCE MRI (9/30 phase). In the early phase of conventional DCE MRI (a), the lesion overlaps with the marked BPE, making it difficult to detect the lesion. In ultrafast DCE MRI, lesion detection is easy because the BPE is weak (b). BPE, background parenchymal enhancement; DCE, dynamic contrast-enhanced.

Ultrafast DCE MRI for evaluating and predicting treatment responses after neoadjuvant therapies

NAC and neoadjuvant systemic therapy (NST) are now standard treatments for locally advanced breast cancer, particularly for HER2-positive and TN breast cancer. It is important to assess the effectiveness or to predict the treatment response of neoadjuvant therapy using imaging modalities. Several studies reported that ultrafast DCE MRI is useful for evaluating treatment response after NAC or NST. Kato et al. showed that a visual analysis at the delayed phase of conventional DCE-MRI showed the highest sensitivity (90%), whereas ultrafast DCE-MRI showed the highest PPV (92%) for evaluating residual tumor after NAC.¹¹¹ They also reported that there were no significant differences between the diameters in ultrafast DCE-MRI and the pathological residual invasive ductal carcinoma (IDC).¹¹¹ Honda et al. reported that a visual analysis of ultrafast DCE MRI for residual tumor after NST provided a higher diagnostic AUC (0.86–0.88) compared to conventional DCE MRI (AUC: 0.68–0.80), and the size differences between the results of pathology and ultrafast DCE MRI were significantly smaller than those between the results of pathology and conventional DCE MRI.⁹² These studies demonstrated that the use of ultrafast DCE MRI provides a diagnostic performance that is equivalent to or superior to that of conventional DCE MRI in assessments of the efficacy of NAC or NST for breast cancer.

Several groups have reported that pretreatment ultrafast DCE MRI is also useful for predicting the pathologic response after NAC. Kim et al. revealed that a higher volume ratio of U1/U2 (U1: the time point at which the lesion starts to enhance, U2: a time point after U1) derived from ultrafast DCE-MRI was independently associated with the pCR attained by patients with TN breast cancer.⁷⁸ Ramtohul and colleagues reported that a WIS value >1.6%/sec was associated with higher pCR rates in HER2-positive breast cancer, whereas a WIS > 1.6%/sec was associated with a low residual cancer burden in luminal HER2-negative and TN breast cancer.⁷⁷ Cao et al. observed that early ultrafast DCE MRI parameter changes are useful for predicting pCR, and they reported that the AUC (0.92) of the model incorporating both ultrafast DCE-MRI parameter change values (from timepoints 1 to 2) and clinicopathologic characteristics was greater than that of the clinical model (0.79) and a model incorporating ultrafast DCE-MRI parameters at timepoint 2 and clinicopathologic characteristics (0.82).⁹³ Kinetic parameters of ultrafast MRI at contralateral BPE were also shown to be useful for predicting pCR.^113,114

Ultrafast DCE MRI as preoperative MRI

Ultrafast DCE MRI is also useful at the preoperative stage. Several investigators have indicated that ultrafast DCE MRI is useful for diagnosing MRI-detected lesions, diagnosing extensive intraductal components, and diagnosing axillary lymph node metastasis.^71,72,83,119 Lee at al. reported that adding use of ultrafast DCE MRI (using MS and initial enhancement phase) improved the specificity of conventional DCE MRI without loss of sensitivity for differentiating MRI-detected lesions in preoperative breast cancer patients.⁷¹ Kim et al. also reported that combination of ultrafast DCE MRI parameters (MS, TTE, and TTP) and clinical, pathological and conventional DCE MRI improved the diagnostic performance of the differentiating MRI-detected lesions.¹¹⁹ Luo et al. reported that analysis of kinetic intra- and peri-tumoral heterogeneity on ultrafast DCE MRI was useful for prediction of the extensive intraductal component of breast cancer.⁸³ In addition, Kikuchi et al. reported the usefulness of ultrafast DCE MRI in diagnosing axillary lymph node metastasis in case report.⁷²

Morphological assessment on ultrafast DCE MRI

Most of the studies of ultrafast MRI have evaluated kinetic parameters; a morphological evaluation was not performed because the spatial resolution was slightly lower than that of conventional MRI in the past. However, Ohashi et al. compared morphological evaluations using ultrafast DCE MRI with those employing conventional MRI, and they found that although ultrafast DCE MRI more often showed a circumscribed margin and homogeneous enhancement compared to conventional DCE MRI, approx. 70% of the mass lesions showed the same morphological assessment between ultrafast and conventional DCE MRI.⁹⁶ They concluded that ultrafast DCE MRI may replace conventional DCE MRI for the collection of morphological information of mass lesions.⁹⁶

In contrast, a study by Kataoka et al. revealed that clustered ring enhancement in NME is less frequent in ultrafast DCE MRI than in DCE MRI.⁵⁸ Few reports of morphological evaluations using ultrafast DCE MRI are available, and further research is thus necessary. A comparison of ultrafast and conventional DCE MRI in representative cases is provided as Fig. 5.

Fig. 5.

A 45 year-old woman with invasive carcinoma of no special type at the right breast (a, b) and a 57 year-old woman with DCIS at the left breast (c, d). (a, c) The early phase of conventional DCE MRI, and (b, d) ultrafast DCE MRI (20/30 phase). In the mass lesion, an irregular to spiculated margin and rim enhancement was revealed by both conventional and ultrafast DCE MRI. In the non-mass enhancement, the lesion shows clustered ring enhancement (suggesting DCIS) on conventional DCE MRI, but the lesion shows clumped enhancement (not clustered ring enhancement) on ultrafast DCE MRI. DCE, dynamic contrast-enhanced; DCIS, ductal carcinoma in situ.

Conclusion

We have provided an overview and a summary of the research concerning abbreviated MRI and ultrafast DCE MRI. Both of these techniques can be performed within a short time, and abbreviated MRI in particular is expected to be useful for lesion detection and differentiating benign and malignant lesions in breast cancer screening. Since ultrafast DCE MRI allows for the evaluation of kinetics, it is useful not only for distinguishing benign from malignant lesions, but also for detecting lesions in cases with strong BPE, predicting the prognostic marker status of breast cancer, assessing the effectiveness of neoadjuvant therapy, examining MRI-detected lesions before surgery, and morphological assessments of mass lesions. It is hoped that both abbreviated MRI and ultrafast DCE MRI will become more widely available.