1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
| import numpy as np
class MultimodalHeartRateFusion:
"""
多模态心率融合器
策略: 基于信号质量指数(SQI)的加权融合
劣质信号自动降权，优质信号自动升权
"""
def __init__(self, fs: int = 30):
self.fs = fs
self.modalities = ['ecg_sw', 'ppg_sw', 'rppg_rgb', 'rppg_ir', 'rppg_thermal']
@staticmethod
def compute_sqi(signal: np.ndarray, fs: int = 30) -> float:
"""
信号质量指数
综合:
1. 信号方差（运动过大→低质量）
2. 频谱集中度（心率频段能量占比）
3. 自相关峰值（周期性强度）
"""
var = np.var(signal)
var_score = np.exp(-var / (np.max(signal)**2))
from scipy.signal import welch
freqs, psd = welch(signal, fs=fs, nperseg=fs*4)
hr_band = (freqs >= 0.7) & (freqs <= 3.0)
hr_power = np.sum(psd[hr_band])
total_power = np.sum(psd) + 1e-8
spectral_score = hr_power / total_power
autocorr = np.correlate(signal, signal, mode='full')
autocorr = autocorr[len(autocorr)//2:]
autocorr /= autocorr[0]
peaks = []
for i in range(1, len(autocorr)-1):
if autocorr[i] > autocorr[i-1] and autocorr[i] > autocorr[i+1]:
peaks.append((i, autocorr[i]))
if peaks:
peak_score = max(p[1] for p in peaks)
else:
peak_score = 0
sqi = 0.3 * var_score + 0.4 * spectral_score + 0.3 * peak_score
return float(np.clip(sqi, 0, 1))
def estimate_hr(self, signal: np.ndarray, fs: int = 30) -> float:
"""单模态心率估计"""
freqs, psd = welch(signal, fs=fs, nperseg=fs*4)
hr_band = (freqs >= 0.7) & (freqs <= 3.0)
if np.any(hr_band):
peak_idx = np.argmax(psd[hr_band])
hr_hz = freqs[hr_band][peak_idx]
return hr_hz * 60
return 0.0
def fuse(
self,
signals: dict,
sample_rates: dict
) -> dict:
"""
多模态心率融合
Args:
signals: {'ecg_sw': array, 'ppg_sw': array, ...}
sample_rates: {'ecg_sw': 500, 'ppg_sw': 100, ...}
Returns:
result: {'hr': float, 'quality': float, 'source': str}
"""
estimates = {}
qualities = {}
for mod, sig in signals.items():
if sig is None or len(sig) == 0:
continue
fs = sample_rates.get(mod, 30)
hr = self.estimate_hr(sig, fs)
sqi = self.compute_sqi(sig, fs)
estimates[mod] = hr
qualities[mod] = sqi
if not estimates:
return {'hr': 0, 'quality': 0, 'source': 'none'}
total_q = sum(qualities.values())
if total_q == 0:
return {'hr': 0, 'quality': 0, 'source': 'none'}
fused_hr = sum(
estimates[m] * qualities[m] for m in estimates
) / total_q
best_mod = max(qualities, key=qualities.get)
return {
'hr': float(fused_hr),
'quality': float(total_q / len(estimates)),
'best_source': best_mod,
'all_estimates': estimates,
'all_qualities': qualities
}
if __name__ == "__main__":
rng = np.random.default_rng(42)
fusion = MultimodalHeartRateFusion()
t = np.arange(0, 10, 1/500)
hr_freq = 1.2
ecg = np.sin(2*np.pi*hr_freq*t) + 0.1*rng.normal(size=len(t))
t_ppg = np.arange(0, 10, 1/100)
ppg = np.sin(2*np.pi*hr_freq*t_ppg) + 0.2*rng.normal(size=len(t_ppg))
t_rppg = np.arange(0, 10, 1/30)
rppg_rgb = 0.5*np.sin(2*np.pi*hr_freq*t_rppg) + 0.5*rng.normal(size=len(t_rppg))
rppg_ir = 0.7*np.sin(2*np.pi*hr_freq*t_rppg) + 0.3*rng.normal(size=len(t_rppg))
rppg_thermal = 0.3*np.sin(2*np.pi*hr_freq*t_rppg) + 0.7*rng.normal(size=len(t_rppg))
signals = {
'ecg_sw': ecg, 'ppg_sw': ppg,
'rppg_rgb': rppg_rgb, 'rppg_ir': rppg_ir,
'rppg_thermal': rppg_thermal
}
sample_rates = {'ecg_sw': 500, 'ppg_sw': 100, 'rppg_rgb': 30, 'rppg_ir': 30, 'rppg_thermal': 30}
result = fusion.fuse(signals, sample_rates)
print("=== 多模态心率融合结果 ===")
print(f"融合心率: {result['hr']:.1f} bpm (真值: 72.0)")
print(f"最佳模态: {result['best_source']}")
print(f"总体质量: {result['quality']:.3f}")
print("\n各模态估计:")
for mod, hr in result['all_estimates'].items():
q = result['all_qualities'][mod]
print(f" {mod:15s}: {hr:6.1f} bpm (SQI: {q:.3f})")
|