Magnetic resonance imaging in identifying dysthyroid optic neuropathy: a systematic review

Figures (3) Tables (2)

Figure 1.
Schematic overview of current MRI modalities and their applications in the identification of dysthyroid optic neuropathy. This figure was created with BioGDP.com^[41].
Figure 2.
PRISMA flow diagram.
Figure 3.
Summary of QUADAS-2 quality assessment findings regarding the risk of bias and applicability concerns across all included studies.

Study/MRI sequences	Participant	Sample size (patients/ eyes)	Age (year)	Sex (male/ female)	TED duration (months)	CAS	Smoking history (yes/no)	¹³¹I therapy (yes/no)	TRAb (IU/L)	BCVA	VFI	MD (dB)	PSD (dB)	IOP (mmHg)	Proptosis (mm)
Wu et al.^[31] (2021) IDEAL-T2WI	DON	30/54	50.5 ± 9.99	18/12	6 (4, 12)	4 (2, 5)^a	11/19	/	/	0.55 ± 0.3^a	/	−4.50 (−10.07, −2.68)^a	3.79 (2.02, 6.57)^a	20.96 ± 4.60^a	19.87 ± 2.59^a
	nDON	30/60	47.63 ± 8.77	13/17	5 (2, 6)	2 (1, 3)^a	11/19	/	/	0.82 ± 0.27^a	/	−2.05 (−2.85, −0.86)^a	1.68 (1.43, 2.13)^a	20.08 ± 6.07^a	19.15 ± 2.86^a
	HC	19/38	46.53 ± 11.97	7/12	/	/	8/11	/	/	0.88 ± 0.32^a	/	/	/	/	/
Liu et al.^[30] (2022) DTI	DON	23/46	55.35 ± 8.89	13/10	6 (2, 10)	2 (1, 4)^a	10/13	3/20	16.86 ± 11.16	0.8 (0.5, 1.0)^a	/	−5.64 ± 7.3^a	5.44 ± 3.88^a	18 (17, 20)^a	19 (16, 21)^a
Liu et al.^[30] (2022) DTI	nDON	36/72	52.14 ± 7.93	18/18	6 (3, 12)	2 (1, 3)^a	14/22	8/28	10.56 ± 11.81	1 (0.8, 1)^a	/	−4.69 ± 1.9^a	4.59 ± 1.69^a	19 (17, 21)^a	18 (16.225, 20)^a
Wu et al.^[29] (2022) IDEAL-T2WI	Training cohort
	DON	46/83	50.93 ± 8.39	25/21	5.59 ± 3.77	4 (2, 5)^a	18/28	/	/	0.6 (0.5, 1.0)^a	/	/	/	/	19 (18, 21)^a
	nDON	42/80	45.90 ± 8.64	19/23	8.93 ± 10.90	2 (1, 2.75)^a	15/27	/	/	0.85 (0.8, 1.0)^a	/	/	/	/	19 (17.25, 20)^a
	Validation cohort
	DON	13/21	52.57 ± 12.17	7/6	6.29 ± 4.33	3 (2, 3)^a	5/8	/	/	0.5 (0.25, 0.8)^a	/	/	/	/	19 (18, 20)^a
	nDON	27/51	47.1 ± 9.0	18/9	6.97 ± 4.9	2 (1.5, 2)^a	9/18	/	/	0.8 (0.8, 1.0)^a	/	/	/	/	19 (18, 20)^a
Zou et al.^[28] (2022) T1WI, FS-T2WI, Dixon-T2WI, T2-mapping, 3DT1	DON	26/43	49.9 ± 12.3	3/23	8 (3, 16)	/	20/6	3/23	19.295 (5.097, 41.315)	/	/	/	/	/	/
	nDON	33/66	45.1 ± 13.1	16/17	6 (3, 12)	/	14/19	3/30	10.755 (3.620, 20.125)	/	/	/	/	/	/
Zhang et al.^[27] (2022) T1WI, T2WI	DON	16/23	45.04 ± 7.85	9/7	/	/	/	/	/	0.66 ± 0.38^a, *	/	−16.51 ± 5.81^a	5.38 ± 3.49^a	21.56 ± 5.01^a	18.39 ± 3.14^a
	nDON	23/37	46.05 ± 5.39	12/11	/	/	/	/	/	0.10 ± 0.16^a, *	/	−1.96 ± 1.70^a	2.13 ± 0.98^a	15.68 ± 2.83^a	17.56 ± 3.26^a
	HC	11/21	43.09 ± 9.09	6/5	/	/	/	/	/	−0.01 ± 0.05^a, *	/	/	/	14.86 ± 2.43^a	/
Song et al.^[26] (2023) T1WI, Dixon-T2WI, 3DT1	DON	24/48	51.00 ± 12.91	13/11	6.00 (3, 12)	4 (4, 5)^a	6/18	/	6.240 (3.060, 25.430)	/	/	/	/	20 (17, 23)^a	18 (14, 22)^a
Song et al.^[26] (2023) T1WI, Dixon-T2WI, 3DT1	nDON	39/78	39.97 ± 10.96	21/18	5.00 (3, 12)	3 (1, 4)^a	12/27	/	7.850 (2.470, 24.880)	/	/	/	/	17 (16, 19)^a	17 (16, 19)^a
Wu et al.^[25] (2024) IDEAL-T2WI	Training cohort
	DON	30/54	50.5±9.99	18/12	6 (4, 12)	4 (2, 5)^a	11/19	4/26	/	0.55 ± 0.3^a	/	/	/	20.96 ± 4.60^a	19.87 ± 2.59^a
	nDON	30/60	47.63±8.77	13/17	5 (2, 6)	2 (1, 3)^a	11/19	5/25	/	0.82 ± 0.27^a	/	/	/	20.08 ± 6.07^a	19.15 ± 2.86^a
	Validation cohort
	DON	14/26	55.2 ± 8.07	9/5	6 (2, 8)	4 (2, 5)^a	4/10	2/12	/	0.61 ± 0.25^a	/	/	/	19.8 ± 3.29^a	19.57 ± 2.03^a
	nDON	15/30	45.29 ± 5.68	9/6	6 (4, 10)	3 (1, 3)^a	3/12	2/13	/	0.87 ± 0.26^a	/	/	/	19.39 ± 2.79^a	18.92 ± 3.08^a
Moledina et al.^[24] (2025) T1WI, STIR-T2WI, DWI	DON	26/NA	/	5/21	/	3.52 ± 1.72^a	7/19	4/15^#	/	/	/	/	/	/	/
	nDON	516/NA	/	105/369^#	/	0.54 ± 0.80^a	115/332^#	66/374^#	/	/	/	/	/	/	/
Li et al.^[23] (2025) T1WI, FS-T2WI	DON	NA/16	53.43 ± 12.55	/	/	/	/	/	/	0.22 ± 0.14^a, *	/	/	/	20.75 ± 5.97^a	/
	Moderate to severe TED	NA/48	43.22 ± 11.87	/	/	/	/	/	/	0.07 ± 0.09^a, *	/	/	/	20.48 ± 6.14^a	/
	Mild TED	NA/45	42.94 ± 9.94	/	/	/	/	/	/	0.01 ± 0.02^a, *	/	/	/	18.73 ± 3.17^a	/
Zhou et al.^[22] (2025) MTI, FS-T2WI, DWI	Training cohort
	DON	30/46	57.93 ± 10.97	13/17	8 (5, 13)	4 (3, 5)^a	12/18	/	8.06 (2.96, 26.35)	0.6 (0.2, 0.7)^a	0.86 (0.42, 0.96)^a	−6.21 (−19.01, −3.09)^a	3.98 (2.12, 8.21)^a	/	21.37 ± 3.25^a
	nDON	74/148	42.85 ± 13.14	24/50	7 (4, 18)	2 (1, 3)^a	28/46	/	2.17 (0.81, 6.76)	1.0 (1.0, 1.0)^a	0.99 (0.97, 0.99)^a	−1.18 (−2.31, −0.38)^a	1.53 (1.30, 2.08)^a	/	19.57 ± 3.21^a
	Validation cohort
	DON	13/20	54.46 ± 14.09	5/8	7 (4, 14)	4 (3, 5)^a	7:6	/	5.93 (1.69, 16.42)	0.6 (0.4, 0.7)^a	0.83 (0.50, 0.97)^a	−6.70 (−16.50, −2.88)^a	4.71 ± 2.97^a	/	20.78 ± 3.40^a
	nDON	32/64	44.50 ± 11.48	5/27	10 (3, 24)	2 (1, 3)^a	12:20	/	2.39 (0.88, 12.97)	1.0 (1.0, 1.0)^a	0.99 (0.98, 1.00)^a	−0.85 (−1.79, 0.06)^a	1.69 ± 0.73^a	/	18.68 ± 3.20^a
Jiang et al.^[38] (2021) Bold-fMRI	DON	16/NA	54.85 ± 5.02	6/10	/	/	/	/	/	0.55 ± 0.25/ 0.35 ± 0.20^b	/	/	/	/	/
Jiang et al.^[38] (2021) Bold-fMRI	nDON	16/NA	53.09 ± 5.16	6/10	/	/	/	/	/	1.05 ± 0.10/ 1.05 ± 0.10^b	/	/	/	/	/
Liu et al.^[37] (2023) Bold-fMRI	DON	20/NA	54.65 ± 7.82	10/10	6 (2, 36)	4 (1~7)/ 4 (1~7)^b	9/11	7/13	17.39 ± 11.69	0.36 ± 0.22/ 0.38 ± 0.17^b	/	/	/	18.92 ± 4.39/ 18.52 ± 3.84^b	19.30 ± 1.90/ 18.85 ± 2.72^b
	nDON	27/NA	52.00 ± 6.63	13/14	8 (3, 33)	1 (0~6)/ 1 (0~6)^b	14/13	9/18	12.53 ± 11.08	0.98 ± 0.14/ 0.91 ± 0.17^b	/	/	/	18.98 ± 6.01/ 20.01 ± 7.27^b	16.80 ± 3.62/ 18.98 ± 6.01^b
	HC	33/NA	54.03 ± 6.72	17/16	/	/	13/20	/	/	/	/	/	/	/	/
Zhou et al.^[36] (2024) Bold-fMRI	DON	28/NA	54.00 ± 11.44	13/15	12.36 ± 8.99	4.14 ± 1.33/ 4.07 ± 1.33^b	12/16	/	/	0.66 ± 0.24/ 0.60 ± 0.27^b	0.83 ± 0.22/ 0.87 ± 0.22^b	−7.32 ± 6.91/ −5.69 ± 6.74^b	3.81 ± 2.75/ 3.39 ± 2.43^b	/	20.14 ± 3.18/ 21.02 ± 3.40^b
	nDON	38/NA	49.18 ± 8.41	15/23	12.61 ± 14.31	2.92 ± 1.22/ 2.94 ± 1.21^b	18/20	/	/	0.97 ± 0.12/ 0.94 ± 0.13^b	/	/	/	/	18.91 ± 3.00/ 19.62 ± 2.92^b
	HC	30/NA	50.30 ± 11.97	11/19	/	/	12/18	/	/	0.98 ± 0.13/ 0.97 ± 0.11^b	/	/	/	/	/
Zhou et al.^[35] (2025) DKI	DON	21/NA	53.7 ± 13.9	9/12	10.0 (6.5, 13.5)	5.0 (3.5, 5.0)/ 5.0 (3.0, 5.0)^b	10/11	/	/	0.80 (0.50, 1.00)/ 0.70 (0.55, 0.80)^b	0.94 (0.80, 0.98)/ 0.93 (0.76, 0.97)^b	/	/	/	19.7 ± 3.4/ 21.2 ± 3.8^b
Zhou et al.^[35] (2025) DKI	nDON	32/NA	45.6 ± 12.7	12/20	8.0 (4.3, 17.3)	2.0 (2.0, 3.0)/ 2.0 (2.0, 3.0)^b	10/22	/	/	1.00 (1.00, 1.00)/ 1.00 (0.80, 1.00)^b	0.99 (0.98, 1.00)/ 0.99 (0.96, 1.00)^b	/	/	/	18.3 ± 2.5/ 18.8 ± 2.5^b
	HC	30/NA	45.4 ± 15.5	10/20	/	/	11/19	/	/	1.00 (1.00, 1.00)/ 1.00 (1.00, 1.00)^b	/	/	/	/	/
Lu et al.^[34] (2025) Bold-fMRI	DON	23/NA	56.04 ± 11.46	11/12	12.04 ± 9.59	4.48 ± 1.27/ 4.39 ± 1.31^b	9/14	/	/	0.60 ± 0.28/ 0.60 ± 0.27^b	0.73 ± 0.31/ 0.82 ± 0.30^b	/	/	/	20.67 ± 3.22/ 21.35 ± 3.53^b
	nDON	34/NA	51.35 ± 7.07	10/24	14.56 ± 14.49	2.21 ± 1.23/ 2.24 ± 1.23^b	11/23	/	/	0.99 ± 0.13/ 0.97 ± 0.14^b	/	/	/	/	17.35 ± 2.79/ 17.62 ± 3.03^b
	HC	27/NA	52.44 ± 7.93	10/17	/	/	13/14	/	/	1.00 ± 0.12/ 1.00 ± 0.11^b	/	/	/	/	/
Zhang et al.^[39] (2024) Dixon-T2WI, Bold-fMRI	DON	47/NA	50.36 ± 12.26	18/29	14.33 (6.00, 20.00)	3.23 ± 1.88^a	12/35	/	11.70 (2.68, 16.60)	0.34 ± 0.23^a	/	/	/	20.00 (17.10, 21.00)^a	19.55 ± 2.88^a
	nDON	56/NA	50.48 ± 9.095	22/34	17.46 (6.00, 15.00)	2.34 ± 1.56^a	13/43	/	4.67 (1.10, 15.58)	0.95 ± 0.08^a	/	/	/	18.50 (17.00, 20.50)^a	19.33 ± 2.58^a
	HC	36/NA	45.64 ± 14.43	16/20	/	/	/	/	/	0.96 ± 0.07^a	/	/	/	/	/
Hu et al.^[40] (2024) T1WI, FS-T2WI, DWI, DKI	DON	25/40	56.6 ± 11.4	11/14	8 (5, 12)	5 (3, 5)^a	10/15	1/24	5.00 (2.30, 11.48)	0.6 (0.4, 0.7)^a	0.88 (0.43, 0.95)	−6.04 (−19.89, −3.40)^a	3.13 (1.98, 5.15)^a	17.7 (14.9, 20.9)^a	20.3 (18.1, 23.8)^a
Hu et al.^[40] (2024) T1WI, FS-T2WI, DWI, DKI	nDON	36/72	45.6 ± 10.7	16/20	7 (3, 18)	2 (1, 3)^a	17/19	3/33	2.14 (1.27, 8.48)	1.0 (1.0, 1.0)^a	0.99 (0.98, 1.00)	−0.92 (−2.19, 0.31)^a	1.50 (1.24, 1.90)^a	15.9 (13.7, 18.8)^a	19.0 (17.0, 22.0)^a
^a Data are presented as the average. ^b Data are presented as right eye (OD)/left eye (OS). *BCVA according to a logMAR visual chart. ^# Existence of missing data. Values in parentheses separated by a comma represent the 25^th and 75^th percentiles, whereas those separated by a tilde represent the range. MRI, magnetic resonance imaging; TED, thyroid eye disease; DON, dysthyroid optic neuropathy; HC, healthy control; CAS, clinical activity score; TRAb, thyrotropin receptor antibody; BCVA, best-corrected visual acuity; VFI, visual field index; MD, mean deviation; PSD, pattern standard deviation; IOP, intraocular pressure; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; DWI, diffusion-weighted imaging; DTI, diffusion tensor imaging; DKI, diffusion kurtosis imaging; MTI, magnetization transfer imaging; Bold-fMRI, blood oxygenation level-dependent functional MRI.

Table 1.

Demographic and clinical characteristics of participants in the included studies.

Study	Parameters or models		Condition (DON vs. nDON)	AUC	Cut-off value	Sensitivity	Specificity	p-value
Wu et al.^[31] (2021)	Optic nerve signal intensity ratio	3 mm behind the eyeball	DON > nDON	0.66	1.71	0.55	0.71	/
	Optic nerve signal intensity ratio	6 mm behind the eyeball	DON > nDON	0.69	1.55	0.57	0.71	/
	Optic nerve water fraction (%)	3 mm behind the eyeball	DON > nDON	0.78	91.55	0.51	0.96	/
	Optic nerve water fraction (%)	6 mm behind the eyeball	DON > nDON	0.76	88.48	0.70	0.70	/
	Optic nerve subarachnoid space (mm)	3 mm behind the eyeball	DON > nDON	0.95	2.35	0.91	0.90	/
	Optic nerve subarachnoid space (mm)	6 mm behind the eyeball	DON > nDON	0.93	2.00	0.91	0.86	/
	Optic nerve sheath diameter (mm)	3 mm behind the eyeball	DON > nDON	0.93	5.45	0.83	0.95	/
	Optic nerve sheath diameter (mm)	6 mm behind the eyeball	DON > nDON	0.91	4.55	0.78	0.87	/
	Optic nerve diameter (mm)	3 mm behind the eyeball	DON > nDON	0.73	2.95	0.72	0.68	/
	Optic nerve diameter (mm)	6 mm behind the eyeball	DON > nDON	0.76	2.35	0.78	0.67	/
	Optic nerve sheath area (mm²)	3 mm behind the eyeball	DON > nDON	0.90	21.5	0.82	0.89	/
	Optic nerve sheath area (mm²)	6 mm behind the eyeball	DON > nDON	0.89	15.5	0.82	0.82	/
	Optic nerve area (6 mm behind the eyeball) (mm²)		DON > nDON	0.73	6.5	0.78	0.62	/
	Optic nerve area/optic nerve sheath area (%)	3 mm behind the eyeball	DON < nDON	0.89	44.09	0.89	0.85	/
	Optic nerve area/optic nerve sheath area (%)	6 mm behind the eyeball	DON < nDON	0.85	45.23	0.89	0.73	/
	Fluid volume in the optic nerve sheath (mm³)		DON > nDON	0.93	0.166	0.85	0.93	/
Liu et al.^[30] (2022)	Optic nerve axial diffusivity (ONAD) (×10^–3 mm²/s)		DON > nDON	0.644	1.927	0.7174	0.6071	/
	Optic nerve radial diffusivity (ONRD) (×10^–3 mm²/s)		DON > nDON	0.640	0.791	0.8696	0.4554	/
	Optic nerve mean diffusivity (ONMD) (×10^–3 mm²/s)		DON > nDON	0.727	1.164	0.9565	0.4732	/
	Optic nerve fractional anisotropy (ONFA)		DON < nDON	0.709	0.574	0.9565	0.3571	/
	ONAD + ONRD		/	0.666	0.711	0.7609	0.5714	/
	ONAD + ONMD		/	0.720	0.719	0.8261	0.5804	/
	ONAD + ONFA		/	0.748	0.728	0.7609	0.6161	/
	ONRD + ONMD		/	0.746	0.716	0.8043	0.6071	/
	ONRD + ONFA		/	0.717	0.713	0.7174	0.6250	/
	ONMD + ONFA		/	0.768	0.738	0.8478	0.6429	/
	ONAD + ONRD + ONMD		/	0.744	0.721	0.8261	0.6071	/
	ONAD + ONRD + ONFA		/	0.748	0.736	0.8043	0.5893	/
	ONAD + ONMD + ONFA		/	0.765	0.709	0.8043	0.6786	/
	ONRD + ONMD + ONFA		/	0.784	0.727	0.7826	0.6786	/
	ONAD + ONRD + ONMD + ONFA		/	0.801	0.729	0.8043	0.6696	/
Wu et al.^[29] (2022)	Training cohort	Convention MRI evaluation (apical crowding sign + optic nerve stretching sign)	/	0.801	/	0.867	0.638	/
		Radiomics signature*	/	0.889	/	0.687	0.950	/
		Radiomics nomogram (conventional MRI evaluation + radiomics signature*)	/	0.921	/	0.867	0.825	/
	Validation cohort	Convention MRI evaluation (apical crowding sign + optic nerve stretching sign)	/	0.745	/	0.762	0.608	0.04 (vs. Radiomics nomogram)
		Radiomics signature*	/	0.848	/	0.667	0.843	/
		Radiomics nomogram (conventional MRI evaluation + radiomics signature*)	/	0.876	/	0.905	0.706	/
Zou et al.^[28] (2022)	Modified muscle index at the posterior globe of 21 mm (mMI at 21 mm) (%)		DON > nDON	0.848	55.1	0.651	0.907	/
	Optic nerve T2 value (ms)		DON > nDON	0.805	119.2	0.884	0.653	/
	mMI at 21 mm + optic nerve T2 value		/	0.918	0.2	0.953	0.760	/
Zhang et al.^[27] (2022)	Extraocular muscle volume (mL)		DON > nDON	0.850	4.035	1.000	0.703	/
	Fat volume (mL)		DON > nDON	0.674	/	/	/	/
	Whole volume (mL)		DON > nDON	0.726	/	/	/	/
	Extraocular muscle volume/fat volume		DON > nDON	0.712	/	/	/	/
Song et al.^[26] (2023)	Optic nerve signal intensity ratio	3 mm behind the eyeball (ONSIR3)	DON > nDON	0.834	1.639	0.829	0.714	/
		6 mm behind the eyeball	DON > nDON	0.732	/	/	/	< 0.001 (vs. ONSIR3)
		9 mm behind the eyeball	DON > nDON	0.641	/	/	/	< 0.001 (vs. ONSIR3)
	Extraocular muscle volume index at the orbital apex (AMI) (%)		DON > nDON	0.898	41.781	0.886	0.890	/
	Extraocular muscle volume at the orbital apex (cm³)		DON > nDON	0.836	/	/	/	0.016 (vs. AMI)
	Extraocular muscle volume (cm³)		DON > nDON	0.802	/	/	/	0.007 (vs. AMI)
	Medial rectus muscle volume (cm³)		DON > nDON	0.806	/	/	/	0.009 (vs. AMI)
	Extraocular muscle volume index (%)		DON > nDON	0.795	/	/	/	0.002 (vs. AMI)
	Barrett index (%)		DON > nDON	0.846	/	/	/	0.048 (vs. AMI)
	Effective orbital volume (cm³)		DON > nDON	0.621	/	/	/	/
	Axial length of the eyeball (mm)		DON < nDON	0.629	/	/	/	/
	ONSIR3 + AMI		/	0.943	/	/	/	0.003 (vs. ONSIR3) 0.013 (vs. AMI)
Wu et al.^[25] (2024)	Training cohort	Cerebrospinal fluid volume in the optic nerve sheath + optic nerve subarachnoid space (3 mm behind the eyeball)	/	0.994	/	0.944	0.967	/
Wu et al.^[25] (2024)	Validation cohort	Cerebrospinal fluid volume in the optic nerve sheath + optic nerve subarachnoid space (3 mm behind the eyeball)	/	0.960	/	0.889	0.893	/
Moledina et al.^[24] (2025)	Apparent diffusion coefficient value of the most involved muscle (×10^–6 mm²/s)		DON > nDON	0.844	1115	0.789	0.782	/
					796.5	1.000	0.248	/
					1498	0.316	0.977	/
Li et al.^[23] (2025)	Volume of the medial orbital wall (cm³)		DON < nDON	0.808	–0.065	0.813	0.753	/
Zhou et al.^[22] (2025)	Training cohort	Mean magnetization transfer ratio of optic nerve (MTRmean) (%)	DON < nDON	0.831	40.0	0.739	0.791	< 0.001 (vs. MTRmean + ADCmean + mMI)
		Mean apparent diffusion coefficient of optic nerve (ADCmean) (×10^-6 mm²/s)	DON < nDON	0.738	1050.0	0.565	0.818	< 0.001 (vs. MTRmean + ADCmean + mMI)
		mMI (%)	DON > nDON	0.862	54.1	0.870	0.750	< 0.001 (vs. MTRmean + ADCmean + mMI)
		MTRmean + ADCmean + mMI	/	0.946	0.154	0.935	0.824	/
	Validation cohort	MTRmean (%)	DON < nDON	0.833	40.0	0.650	0.813	0.028 (vs. MTRmean + ADCmean + mMI)
		ADCmean (×10^-6 mm²/s)	DON < nDON	0.768	1050.0	0.550	0.813	0.005 (vs. MTRmean + ADCmean + mMI)
		mMI (%)	DON > nDON	0.852	54.1	0.750	0.797	0.058 (vs. MTRmean + ADCmean + mMI)
		MTRmean + ADCmean + mMI	/	0.926	0.154	0.750	0.844	/
Jiang et al.^[38] (2021)	Regional homogeneity (ReHo) value of corpus callosum/cingulate gyrus		DON < nDON	0.933	/	/	/	/
Jiang et al.^[38] (2021)	ReHo value of parietal lobe/middle frontal gyrus		DON < nDON	0.938	/	/	/	/
Liu et al.^[37] (2023)	ReHo value of left insula		DON < nDON	0.757	/	/	/	/
Liu et al.^[37] (2023)	ReHo value of left posterior cingulate cortex		DON > nDON	0.843	/	/	/	/
	ReHo value of right superior temporal lobe		DON < nDON	0.646	/	/	/	/
	ReHo value of left insula + ReHo value of left posterior cingulate cortex		/	0.915	/	/	/	/
	ReHo value of left insula + ReHo value of right superior temporal lobe		/	0.763	/	/	/	/
	ReHo value of left posterior cingulate cortex + ReHo value of right superior temporal lobe		/	0.852	/	/	/	/
	ReHo value of left insula + ReHo value of left posterior cingulate cortex + ReHo value of right superior temporal lobe		/	0.915	/	/	/	/
Zhou et al.^[36] (2024)	Model of static functional network (σ + degree centrality in left medial orbital part of superior frontal gyrus + degree centrality in left middle occipital gyrus + degree centrality in right inferior occipital gyrus + nodal efficiency in left medial orbital part of superior frontal gyrus + nodal efficiency in right medial orbital part of superior frontal gyrus + nodal efficiency in left gyrus rectus + nodal efficiency in right inferior occipital gyrus)		/	0.793	/	/	/	/
	Model of dynamic functional network (variance of global efficiency + variance of characteristic path length)		/	0.766	/	/	/	/
	Model of static functional network + model of dynamic functional network		/	0.836	/	/	/	/
Zhou et al.^[35] (2025)	Radial kurtosis value of right posterior thalamic radiation (include optic radiation)		DON < nDON	0.838	/	/	/	/
	Radial kurtosis value of left posterior thalamic radiation (include optic radiation)		DON < nDON	0.802	/	/	/	/
	Mean kurtosis value of right posterior thalamic radiation (include optic radiation)		DON < nDON	0.795	/	/	/	/
	Radial kurtosis value of left sagittal stratum		DON < nDON	0.786	/	/	/	/
	Radial kurtosis value of right sagittal stratum		DON < nDON	0.753	/	/	/	/
	Axial kurtosis value of right posterior corona radiata		DON < nDON	0.751	/	/	/	/
	Radial kurtosis value of left retrolenticular part of internal capsule		DON < nDON	0.723	/	/	/	/
Lu et al.^[34] (2025)	Static amplitude of low-frequency fluctuation (ALFF) model (static ALFF value of the left lingual gyrus + static ALFF value of the right lingual gyrus + static ALFF value of the right cuneus + static ALFF value of the left medial part of the superior frontal gyrus + static ALFF value of the right medial part of the superior frontal gyrus + static ALFF value of the right dorsolateral part of the superior frontal gyrus + static ALFF value of the right precentral gyrus)		/	0.8440	/	0.7826	0.9118	0.0155 (vs. Dynamic ALFF model)
	Dynamic ALFF model (dynamic ALFF value of the left lingual gyrus + dynamic ALFF value of the right cuneus + dynamic ALFF value of the right orbital part of the middle frontal gyrus + dynamic ALFF value of the right dorsolateral part of the superior frontal gyrus)		/	0.9987	/	1.0000	0.9706	/
	Static ReHo model (static value of the right lingual gyrus + static value of the right supramarginal gyrus)		/	0.7762	/	0.7391	0.8235	0.0007 (vs. Dynamic ALFF model)
	Dynamic ReHo model (dynamic value of the right supramarginal gyrus)		/	0.7609	/	0.6957	0.7647	0.0004 (vs. Dynamic ALFF model)
Zhang et al.^[39] (2024)	ALFF value of the right superior frontal gyrus, orbital part		DON > nDON	0.64	/	/	/	/
	ALFF value of right hippocampus		DON > nDON	0.66	/	/	/	/
	ALFF value of right superior temporal gyrus		DON > nDON	0.66	/	/	/	/
	Posterior mMI		DON > nDON	0.76	/	/	/	0.1456 (vs. Combined model)
	Combined model (ALFF value of the right hippocampus + ALFF value of the right superior temporal gyrus + posterior mMI)		/	0.80	/	/	/	/
Hu et al.^[40] (2024)	Apical muscle index (MI)		DON > nDON	0.822	0.548	0.750	0.806	/
	Optic nerve apical area index (ONAI)		DON > nDON	0.791	0.518	0.700	0.819	0.588 (vs. MI)
	Optic nerve apparent diffusion coefficient (ONADC) (×10^–3 mm²/s)		DON < nDON	0.719	1.123	0.725	0.653	0.098 (vs. MI)
	Optic nerve signal intensity ratio		DON > nDON	0.619	1.090	0.400	0.847	0.003 (vs. MI)
	Kurtosis fractional anisotropy value of optic tract (OTKFA)		DON < nDON	0.745	0.412	0.400	0.972	0.080 (vs. ORKFA)
	Kurtosis fractional anisotropy value of optic radiation (ORKFA)		DON < nDON	0.827	0.421	0.600	1.000	/
	Kurtosis fractional anisotropy value of Brodmann Area 17 (BA17KFA)		DON < nDON	0.789	0.283	0.725	0.833	0.261 (vs. ORKFA)
	Kurtosis fractional anisotropy value of Brodmann Area 18		DON < nDON	0.698	0.286	0.575	0.833	0.007 (vs. ORKFA)
	Kurtosis fractional anisotropy value of Brodmann Area 19		DON < nDON	0.718	0.315	0.475	0.972	0.011 (vs. ORKFA)
	Mean kurtosis of optic tract		DON < nDON	0.625	0.900	0.450	0.917	0.002 (vs. ORKFA)
	Mean kurtosis of optic radiation		DON < nDON	0.597	0.852	0.425	0.944	< 0.001 (vs. ORKFA)
	Orbital MRI model (MI + ONAI + ONADC)		/	0.890	/	0.775	0.903	0.257 (vs. visual pathway DKI model))
	Visual pathway DKI model (OTKFA + ORKFA + BA17KFA)		/	0.832	/	0.700	0.917	0.005 (vs. combined model)
	Combined model (orbital MRI model + visual pathway DKI model)		/	0.926	/	0.800	0.944	0.177 (vs. orbital MRI model)
* Radiomics signature: Rad-score = 0.046222 + (original_shape_LeastAxisLength × 0.295835) – (wavelet.HHH_glrlm_LowGrayLevelRunEmphasis × 0.168698) + (wavelet.HHH_glszm_SizeZoneNonUniformityNormalized × 0.078249) – (original_glcm_InverseVariance × 0.215994) + (wavelet.HLH_glszm_LargeAreaHighGrayLevelEmphasis × 0.256508) – (wavelet.HHH_glcm_ClusterShade × 0.061405) – (original_glrlm_ShortRunLowGrayLevel Emphasis × 0.238402) + (original_gldm_LargeDependenceHighGrayLevelEmphasis × 0.006713). DON, dysthyroid optic neuropathy; AUC, area under the receiver operating characteristic (ROC) curve; ONAD, optic nerve axial diffusivity; ONRD, optic nerve radial diffusivity; ONMD, optic nerve mean diffusivity; ONFA, optic nerve fractional anisotropy; MRI, magnetic resonance imaging; mMI, modified muscle index; ONSIR3, optic nerve signal intensity ratio at 3 mm behind the eyeball; AMI, extraocular muscle volume index at the orbital apex; MTRmean, mean magnetization transfer ratio; ADCmean, mean apparent diffusion coefficient; ReHo, regional homogeneity; ALFF, amplitude of low-frequency fluctuation; MI, apical muscle index; ONAI, optic nerve apical area index; ONADC, optic nerve apparent diffusion coefficient; OTKFA, kurtosis fractional anisotropy value of optic tract; ORKFA, kurtosis fractional anisotropy value of optic radiation; BA17KFA, kurtosis fractional anisotropy value of Brodmann Area 17; DKI, diffusion kurtosis imaging.

Table 2.

The main findings of MRI studies on the diagnosis of DON.