Left atrial volume quantification in cardiac MRI: comparison of different approaches

Background: The left atrium (LA) is increasingly recognized as a key marker of cardiovascular health, providing prognostic information related to diastolic dysfunction[1-3] and atrial fibrillation[4]. The reference method is to acquire LA in a short-axis (SAX) stack (Simpson’s), which requires complete atrial coverage and is not routinely used. Alternative approaches, including area-length and 2- and 4-chamber Simpson’s methods, are widely used but prone to limitations such as LA foreshortening and LA volume underestimation[5]. Others have proposed on-axis views of the LA, which also requires additional acquisition, and are not widely adopted[6]. As LA basal area is usually obtained on the short-axis LV stack and LA length can be measured from 2ch and 4ch views, this study aimed to evaluate SAX LA area-length method as an alternative method to SAX Simpon’s.

Methods: Adult subjects were prospectively enrolled. SSFP real-time cines were acquired in the 2,4 and 3 chamber and short-axis views covering both ventricles and atria on a 3T scanner (Magnetom Vida, Siemens) . LA length was used as either the average of 2 and 4-chamber values (length_ave) or the larger of the two lengths (length_max). Planimetered LA areas were measured from 2 and 4-chamber views (Area_2ch and Area_4ch) as well as from basal SAX view (Area_SAX) (Figure 1). The LA appendage and pulmonary veins were excluded from all volumetric analyses.

LA volumes were analyzed using 4 different approaches:

(1) Simpson’s methods on dedicated LA short-axis stack (V _Simpson-SAX) [7]; (2) Biplane area-length method V_AL= 0.85 x (Area_2chx Area_4ch)/length_ave[8]; (3) Two and 4-chamber biplane Simpson’s methods (V _Simpson-LAX); (4) SAX LA area-length methods V_{SAX AL} = Area_{SAX basal LA} x length_max.

Robust linear regression (Huber M-estimator), Spearman rank correlation, Bland-Altman analysis and repeated measures ANOVA were performed to evaluate method comparisons.

Results: A total of 13 subjects (59±13 years, ranging from 26 to 80 years, male: 7/13, 54%) were included. Comparisons of V _Simpson-LAX, V_AL, and V_{SAX AL} with the gold-standard of V _Simpson-SAX are summarized in Table 1. Robust regression indicated strong linear associations between all three methods and V _Simpson-SAX (robust r > 0.9 for each). However, V_{SAX AL} preserved patient ranking well, with rho=0.791, p=0.001, whereas the rank order of individual patients by V _Simpson-LAX and V_AL was less consistent with the reference test (Table 1 and Figure 2). In addition, V_{SAX AL} exhibited the smallest mean absolute difference from V _Simpson-SAX of 13ml ±3 ml(Table 1 and Figure 2).

Conclusion: Among the different approaches of LA measurement, values derived from the combination of longitudinal and a single LA view at the base of the SAX stack (V_{SAX AL}) demonstrated a good correlation with the reference method V _Simpson-SAX with a small absolute difference, and preserved patient ranking.

Table 1: Comparison between V Simpson-SAX and other different methods of LA volume measurement

Methods	Robust r value	Spearman rho value	P value	Difference from V Simpson-SAX (mean±SD, ml)
V Simpson-LAX	0.904	0.377	0.204	17±5
VAL	0.922	0.423	0.150	15±3
VSAX AL	0.952	0.791	0.001	13±3

Figure 1: Examples of LA measurement. A-F: LA V Simpson-SAX was measured on dedicated LA short axis stacks by tracking the LA endocardium. LA appendage and pulmonary veins were excluded from LA volume; G and H: LA volume was measured using area-length method (V AL) or Simpson’s (V Simpson-LAX) from Longitudinal two- (G) and four-chamber views (H). VSAX AL was calculated using the area measured from (A) and longest length from (G) or (H).
Figure 2: Bland-Altman curves between V Simpson-SAX and other methods of LA volume measurement